Fengyun satellite observation data assimilation method and device, electronic equipment and storage medium

Abstract

This invention provides a method, apparatus, electronic device, and storage medium for assimilating Fengyun satellite observation data, relating to the field of meteorological data processing technology. It aligns the spatial grid points of the real-time land surface temperature with the background field using spatiotemporal resolution. If the measured area is a non-missing area, the real-time land surface temperature corresponding to the non-missing area is used to replace the land surface temperature in the background field at the corresponding time. If the measured area is a missing area, the land surface temperature in the background field at the corresponding time is corrected based on a land surface temperature correction model to obtain an optimized background field. The observed radiance is combined with the optimized background field to construct an objective function and calculate the analysis field. The analysis field is iteratively optimized to minimize the objective function, outputting the optimal analysis field. This invention can reduce background field errors, enabling the system to have better satellite data assimilation capabilities, while also improving the assimilation effect of satellite data under complex terrain at high altitudes.

Description

Technical Field

[0001] This invention relates to the field of meteorological data processing technology, and in particular to a method, apparatus, electronic device, and storage medium for assimilating Fengyun satellite observation data. Background Technology

[0002] Weather forecasting has significant social and economic benefits for agriculture, transportation, and power industries. Currently, the main method for weather forecasting is numerical weather prediction, which describes the dynamic and thermodynamic processes of the atmosphere using a system of equations. Under certain initial and boundary conditions, computers solve these equations to obtain forecast values ​​for various atmospheric elements over a future time period. Generally, for short- to medium-term timescales with forecast durations of less than one week, numerical weather prediction is essentially an initial value problem involving differential equations, and the accuracy of the initial field has a decisive impact on the forecast results. Assimilation techniques involve fusing various observational data (such as satellite observations and ground station observations) with numerical models to generate a numerical model of the atmospheric state that more closely approximates reality, providing a more accurate initial field for forecasting. To reduce initial field errors, it is necessary to incorporate as much observational data as possible into the assimilation system.

[0003] In related technologies, the atmospheric radiative transfer equation is parameterized to construct the relationship between atmospheric state variables and satellite-observed radiance. This relationship is then used as an observation operator in the assimilation system to directly assimilate satellite-observed radiance data. As observation operators for spaceborne microwave radiometers, fast radiative transfer models such as RTTOV (Radiative Transfer for TIROS-N Operational Vertical Sounder) and CRTM (Community Radiative Transfer Model) are widely used in advanced assimilation systems such as WRFDA (Weather Research and Forecasting Data Assimilation) and GSI (Gridpoint Statistical Interpolation). Given atmospheric state variables in the assimilation background field, these models can be used to quickly calculate satellite radiative brightness temperature. However, due to its high altitude and complex terrain, the simulation accuracy of land surface temperature and land surface emissivity in radiative transfer models is often low in plateau regions. These are important parameters for calculating atmospheric radiativeness. This means that a large portion of satellite observations over some areas, such as plateaus, cannot be included in the assimilation system, resulting in missing satellite observation information in numerical forecasts. Consequently, the errors in radiative transfer model calculations are relatively large, and the assimilation effect of satellite data under high-altitude and complex terrain is poor. Summary of the Invention

[0004] This invention provides a method, apparatus, electronic device, and storage medium for assimilating Fengyun satellite observation data, which addresses the shortcomings of traditional Fengyun satellite observation data assimilation methods, such as missing satellite observation information, large calculation errors in radiative transfer modes, and poor assimilation effects of satellite data under complex terrain at high altitudes.

[0005] This invention provides a method for assimilating Fengyun satellite observation data, comprising:

[0006] Generate the background field of the numerical model using the region model;

[0007] The real-time land surface temperature is retrieved from the thermal infrared channel of the first satellite observation equipment. The real-time land surface temperature is then aligned with the spatial grid of the background field of the numerical model to obtain the measured area of ​​land surface temperature that matches the spatial grid of the background field.

[0008] If the measured area is a non-missing area, the real-time land surface temperature corresponding to the non-missing area is used to replace the land surface temperature in the background field at the corresponding time; if the measured area is a missing area, the land surface temperature in the background field at the corresponding time is corrected based on the land surface temperature correction model to obtain an optimized background field.

[0009] Based on historical background field data, an error description of each variable in the background field is constructed to obtain background error information. The optimized background field and the background error information are then converted into simulated emissivity.

[0010] Collect raw radiance observation data acquired by the second satellite observation equipment, preprocess the raw radiance observation data, and obtain the observed radiance;

[0011] The difference between the observed radiance and the simulated radiance is calculated to obtain the observation increment;

[0012] The observed radiance is combined with the optimized background field to obtain the analysis field. The analysis field is then iteratively optimized to output the optimal analysis field.

[0013] The land surface temperature correction model is trained based on historical land surface temperatures derived from inversion and meteorological features simulated by regional models for the corresponding time periods.

[0014] According to the Fengyun satellite observation data assimilation method provided by the present invention, the correction of the land surface temperature in the background field at the corresponding time based on the land surface temperature correction model includes:

[0015] Extract feature vectors of meteorological elements closely related to land surface temperature from numerical models;

[0016] Collect land surface temperature data retrieved from historical time periods, as well as meteorological feature data from regional model simulations of the corresponding time periods, and construct a training set;

[0017] The random forest algorithm was used to fit the training samples in the training set to obtain the land surface temperature correction model.

[0018] The meteorological element feature vectors are input into the land surface temperature correction model to obtain the predicted land surface temperature, which is then used as the optimized land surface temperature.

[0019] According to the Fengyun satellite observation data assimilation method provided by the present invention, the preprocessing of the original radiance observation data to obtain the observed radiance includes:

[0020] The raw radiance observation data is filtered;

[0021] The observed radiance is obtained by correcting the bias in the filtered radiance observation data using the dynamic bias correction method.

[0022] According to the Fengyun satellite observation data assimilation method provided by the present invention, the step of constructing an error describing each variable in the background field based on historical background field data to obtain background error information includes:

[0023] Based on historical background field data of the study area, key variables in the pattern background field are extracted;

[0024] Using the background error calculation module in the WRFDA assimilation system, the error values ​​and spatial correlations of the key variables are calculated.

[0025] Based on the error values ​​and spatial correlation, a background error covariance matrix is ​​constructed. The covariance matrix is ​​used to describe the error distribution of each variable in the mode background field and their interrelationships.

[0026] The background error covariance matrix is ​​standardized using the WRFDA assimilation system to generate background error information.

[0027] According to the Fengyun satellite observation data assimilation method provided by the present invention, the step of converting the optimized background field and the background error information into simulated emissivity includes:

[0028] Based on the type of the second satellite observation equipment, match the coefficient file of the radiative transfer mode;

[0029] The coefficient file, the optimized background field, and the background error information are input into the observation operator, which converts the background field variables into simulated emissivity.

[0030] According to the Fengyun satellite observation data assimilation method provided by this invention, after obtaining the observed radiance, it also...

[0031] The difference between the observed radiance and the simulated radiance is calculated to obtain the observation increment.

[0032] Observation points with observation increments greater than a preset threshold on the spatial grid are removed, and the observation radiance used for subsequent assimilation is selected.

[0033] The selected observed radiance is combined with the optimized background field to obtain the analysis field.

[0034] According to the Fengyun satellite observation data assimilation method provided by the present invention, the iterative optimization of the analysis field to output the optimal analysis field includes:

[0035] The selected observed radiance is combined with the optimized background field using a three-dimensional variational method to construct an objective function, and the analysis field is calculated based on the objective function.

[0036] The optimized background field is replaced by the analytical field, and then recombined with the screened observed radiance. The objective function is iteratively calculated to minimize the objective function, thereby obtaining the optimal analytical field.

[0037] The present invention also provides a Fengyun satellite observation data assimilation device, comprising:

[0038] The generation module is used to generate the background field of the numerical model using the regional model;

[0039] The alignment module is used to obtain the real-time land surface temperature from the thermal infrared channel of the first satellite observation equipment, and to align the real-time land surface temperature with the spatial grid of the background field of the numerical model in terms of spatiotemporal resolution, so as to obtain the measured area of ​​land surface temperature that matches the spatial grid of the background field.

[0040] The optimization module is used to replace the land surface temperature in the background field at the corresponding time with the real-time land surface temperature corresponding to the non-missing measurement area if the measured area is a non-missing measurement area; and to correct the land surface temperature in the background field at the corresponding time based on the land surface temperature correction model if the measured area is a non-missing measurement area, so as to obtain an optimized background field.

[0041] The module is used to construct errors describing each variable in the background field based on historical background field data, obtain background error information, and convert the optimized background field and the background error information into simulated emissivity.

[0042] The acquisition module is used to collect raw radiance observation data acquired by the second satellite observation equipment, and to preprocess the raw radiance observation data to obtain the observed radiance.

[0043] The output module is used to combine the observed radiance with the optimized background field to obtain the analysis field, iteratively optimize the analysis field, and output the optimal analysis field.

[0044] The land surface temperature correction model is trained based on historical land surface temperatures derived from inversion and meteorological features simulated by regional models for the corresponding time periods.

[0045] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the Fengyun satellite observation data assimilation method as described in any of the preceding claims.

[0046] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the Fengyun satellite observation data assimilation method described in any of the preceding claims.

[0047] The present invention provides a method, apparatus, electronic device, and storage medium for assimilating Fengyun satellite observation data. This involves generating a numerical model background field using a regional model; obtaining real-time land surface temperature from the thermal infrared channel of a first satellite observation device; aligning the real-time land surface temperature with the spatial grid of the numerical model background field using spatiotemporal resolution to obtain a measured area of ​​land surface temperature matching the spatial grid of the background field; if the measured area is a non-missing area, replacing the land surface temperature in the background field at the corresponding time with the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, correcting the land surface temperature in the background field at the corresponding time based on a land surface temperature correction model to obtain an optimized background field; and constructing an error description of each variable in the background field based on historical background field data to obtain the background temperature. Error information is used to convert the optimized background field and the background error information into simulated radiance; raw radiance observation data collected by the second satellite observation equipment is collected, and the raw radiance observation data is preprocessed to obtain observed radiance; the observed radiance is combined with the optimized background field to obtain an analysis field, and the analysis field is iteratively optimized to output the optimal analysis field; wherein, the land surface temperature correction model is trained based on historically inverted land surface temperature and meteorological element characteristics simulated by regional models for the corresponding time period. This invention reduces the error of the background field by optimizing the land surface temperature parameters of the background field, allowing more satellite microwave observation data to enter the assimilation system, enabling the system to have better satellite data assimilation capabilities, and improving the assimilation effect of satellite data under high-altitude complex terrain. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in this 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 some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0049] Figure 1 This is one of the flowcharts illustrating the Fengyun satellite observation data assimilation method provided in this embodiment of the invention;

[0050] Figure 2 This is the second flowchart illustrating the Fengyun satellite observation data assimilation method provided in this embodiment of the invention.

[0051] Figure 3 This is a schematic diagram of background field land surface temperature correction provided in an embodiment of the present invention;

[0052] Figure 4 This is a schematic diagram of the direct assimilation of data from the window area and near-surface channel of the Fengyun spaceborne microwave humidity meter provided in an embodiment of the present invention;

[0053] Figure 5 This is a schematic diagram showing the assimilation quantity and root mean square error before and after optimization provided in an embodiment of the present invention;

[0054] Figure 6 This is a schematic diagram of the vertical distribution of the root mean square error of the atmospheric temperature profile before and after optimization, provided in an embodiment of the present invention.

[0055] Figure 7 This is a functional structure diagram of the Fengyun satellite observation data assimilation device provided in an embodiment of the present invention;

[0056] Figure 8 This is a functional structure diagram of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0058] Figure 1 A flowchart of the Fengyun satellite observation data assimilation method provided in the embodiments of the present invention is shown below. Figure 1 As shown, the Fengyun satellite observation data assimilation method provided in this embodiment of the invention includes:

[0059] Step 101: Generate the numerical model background field using the region model;

[0060] Step 102: Obtain the real-time land surface temperature from the thermal infrared channel of the first satellite observation equipment, and align the real-time land surface temperature with the spatial grid of the background field of the numerical model to obtain the measured area of ​​land surface temperature that matches the spatial grid of the background field.

[0061] Due to factors such as clouds and precipitation, the Feng-4 land surface temperature product often has a large number of missing values. The missing values ​​are determined by the spatiotemporally matched Feng-4 land surface temperature.

[0062] Step 103: If the measured area is a non-missing area, then the real-time land surface temperature corresponding to the non-missing area is used to replace the land surface temperature in the background field at the corresponding time; if the measured area is a missing area, then the land surface temperature in the background field at the corresponding time is corrected based on the land surface temperature correction model to obtain an optimized background field.

[0063] Step 104: Construct an error description of each variable in the background field based on historical background field data to obtain background error information, and convert the optimized background field and the background error information into simulated emissivity;

[0064] Step 105: Collect the raw radiance observation data acquired by the second satellite observation equipment, and preprocess the raw radiance observation data to obtain the observed radiance;

[0065] Step 106: Combine the observed radiance with the optimized background field to obtain the analysis field, perform iterative optimization on the analysis field, and output the optimal analysis field;

[0066] The land surface temperature correction model is trained based on historical land surface temperatures derived from inversion and meteorological features simulated by regional models for the corresponding time periods.

[0067] Traditional assimilation techniques parameterize the atmospheric radiative transfer equation, construct the relationship between atmospheric state variables and satellite-observed radiance, and incorporate this relationship as an observation operator into the assimilation system to directly assimilate satellite-observed radiance data. Fast radiative transfer models, such as RTTOV and CRTM, are widely used as observation operators in advanced assimilation systems like WRFDA and GSI, given atmospheric state variables in the assimilation background field, to quickly calculate satellite radiative brightness temperature. However, due to their high altitude and complex terrain, the simulation accuracy of land surface temperature and emissivity in high-altitude regions is often low, even though these are crucial parameters for calculating atmospheric radiance. This results in a significant portion of satellite observations over areas like high-altitude plateaus not being incorporated into the assimilation system, leading to missing satellite observation information in numerical weather predictions, substantial errors in radiative transfer model calculations, and poor assimilation of satellite data under high-altitude and complex terrain conditions.

[0068] The Fengyun satellite observation data assimilation method provided in this invention generates a numerical model background field using a regional model; it obtains real-time land surface temperature from the thermal infrared channel of a first satellite observation device, aligns the real-time land surface temperature with the spatial grid of the numerical model background field using spatiotemporal resolution, and obtains a measured area of ​​land surface temperature matching the spatial grid of the background field; if the measured area is a non-missing area, it replaces the land surface temperature in the background field at the corresponding time with the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, it corrects the land surface temperature in the background field at the corresponding time based on a land surface temperature correction model to obtain an optimized background field; and it constructs an error description of each variable in the background field based on historical background field data to obtain background error information. The optimized background field and the background error information are converted into simulated radiance; raw radiance observation data collected by the second satellite observation equipment are collected, and the raw radiance observation data are preprocessed to obtain observed radiance; the observed radiance is combined with the optimized background field to obtain an analysis field, and the analysis field is iteratively optimized to output the optimal analysis field; wherein, the land surface temperature correction model is trained based on historically inverted land surface temperature and meteorological element characteristics simulated by regional models for the corresponding time period. This invention reduces the error of the background field by optimizing the land surface temperature parameters of the background field, allowing more satellite microwave observation data to enter the assimilation system, enabling the system to have better satellite data assimilation capabilities, and improving the assimilation effect of satellite data under high-altitude complex terrain.

[0069] Based on any of the above embodiments, the correction of the land surface temperature in the background field at the corresponding time based on the land surface temperature correction model includes:

[0070] Step 201: Extract meteorological feature vectors closely related to land surface temperature from the numerical model;

[0071] Step 202: Collect land surface temperature data retrieved from historical time periods, as well as regional model simulation meteorological element characteristic data for the corresponding time periods, and construct a training set;

[0072] Step 203: Use the random forest algorithm to fit the training samples in the training set to obtain the land surface temperature correction model;

[0073] Step 204: Input the meteorological element feature vector into the land surface temperature correction model to obtain the predicted land surface temperature, which is then used as the optimized land surface temperature.

[0074] In this embodiment of the invention, the 2-meter temperature and humidity, surface radiation, surface heat flux, wind speed, and land surface temperature, which are closely related to land surface temperature in the forecast model, are used as meteorological feature vectors for subsequent correction of land surface temperature. Historical wind-induced land surface temperature and WRF-simulated meteorological feature data for the corresponding time period are used as the training set. A random forest algorithm is employed to fit the data using the training samples, and the land surface temperature corresponding to the feature vector is used as the prediction result to obtain a machine learning correction model for land surface temperature.

[0075] This invention combines the latest high spatiotemporal resolution and timeliness of wind-based land surface temperature (FST) products with machine learning methods to achieve real-time generation of seamless, high spatiotemporal resolution optimized FST products. Previously, real-time, seamlessly spaced FST observations were lacking in many land areas. Therefore, in operational or research projects implementing direct assimilation of spaceborne microwave observations, it was difficult to correct background FST temperatures in a timely manner, wasting significant spaceborne microwave observation resources. This invention combines the latest high spatiotemporal resolution and timeliness of wind-based FST products with machine learning methods to achieve real-time generation of seamless optimized FST products, directly optimizing and correcting background FST temperatures, and enabling the assimilation of more spaceborne microwave observation data.

[0076] Based on any of the above embodiments, the preprocessing of the original radiance observation data to obtain the observed radiance includes:

[0077] Step 301: Filter the raw emissivity observation data;

[0078] Step 302: Correct the observed radiance data after filtering using the dynamic bias correction method to obtain the observed radiance.

[0079] In this embodiment of the invention, the radiance observation data is decoded and read by the input data preprocessing module, the observation information is extracted, quality control is performed, obviously invalid data and excessively large data such as satellite observation zenith angle are removed, and then the dynamic bias correction method (VARBC) is used to correct the bias of the observation data, thereby obtaining accurate observed radiance.

[0080] Based on any of the above embodiments, the step of constructing an error describing each variable in the background field based on historical background field data to obtain background error information includes:

[0081] Step 401: Extract key variables from the pattern background field based on historical background field data of the study area;

[0082] Step 402: Calculate the error values ​​and spatial correlations of the key variables using the background error calculation module in the WRFDA assimilation system;

[0083] Step 403: Based on the error value and spatial correlation, construct the background error covariance matrix. The covariance matrix is ​​used to describe the error distribution of each variable in the mode background field and their interrelationships.

[0084] Step 404: The background error covariance matrix is ​​standardized using the WRFDA assimilation system to generate background error information.

[0085] Based on any of the above embodiments, the step of converting the optimized background field and the background error information into simulated emissivity includes:

[0086] Step 501: Match the coefficient file of the radiative transfer mode according to the type of the second satellite observation equipment;

[0087] Step 502: Input the coefficient file, the optimized background field, and the background error information into the observation operator, and convert the background field variable into simulated emissivity through the observation operator.

[0088] This invention constructs a background field land surface temperature correction module in the mainstream data assimilation system (WRFDA) and the fast radiative transfer mode (RTTOV). The background field land surface temperature output by the mode is optimized and then fed into the observation operator, enabling the new system to assimilate more window area and near-surface channel data from the Fengyun satellite-borne microwave hygrometer, which can promote the application of land surface information in weather forecasting.

[0089] Traditional methods for adjusting the bias in background field land surface temperature (TST) indirectly involve modifying the parameterization scheme of physical processes related to TST within the model. However, the optimal parameterization scheme for TST varies across different regions, requiring extensive testing and complex model development. Furthermore, changes to the parameterization scheme can alter other variables within the model, complicating the problem. In this invention, the module for optimizing background field TST is independent of other modules. The optimization and correction process does not affect other variables in the model, and the optimization method is applicable to different regions, exhibiting universality. Moreover, the optimization and correction scheme in this invention is primarily based on machine learning, significantly reducing time and computational resources compared to previous methods that required testing different parameterization schemes.

[0090] Based on any of the above embodiments, after obtaining the observed emissivity, the method further includes:

[0091] Step 601: Calculate the difference between the observed radiance and the simulated radiance to obtain the observation increment;

[0092] Step 602: Remove observation points with excessively large observation increments on the spatial grid, select the observation radiance to be used for subsequent assimilation, and combine it with the background field after optimizing the land surface temperature to obtain the analysis field.

[0093] By optimizing the land surface temperature parameter in the observation operator, this invention can effectively reduce the observation increment, improve the background field quality, and allow more satellite radiance observation data to enter the assimilation system.

[0094] Based on any of the above embodiments, the iterative optimization of the analysis field to output the optimal analysis field includes:

[0095] Step 701: Using a three-dimensional variational method, combine the screened observed emissivity with the background field after optimizing the land surface temperature to construct an objective function and thus calculate the analysis field;

[0096] Step 702: Replace the background field after optimizing the land surface temperature with the analysis field, and recombine it with the screened observed radiance. Iteratively calculate the objective function to minimize the objective function and obtain the optimal analysis field.

[0097] Based on any of the above embodiments, such as Figure 2 As shown, the Fengyun satellite observation data assimilation method provided in this embodiment of the invention specifically includes:

[0098] The raw emissivity observation data from the Fengyun-3 microwave hygrometer were collected and organized for subsequent satellite data assimilation.

[0099] The radiance observation data is input into the data preprocessing module for decoding and reading, and its observation information is extracted. Quality control is performed to remove obviously invalid data and excessively large data such as satellite observation zenith angle. The dynamic bias correction method (VARBC) is used to correct the bias of the observation data.

[0100] Based on the type of spaceborne microwave humidity meter, a specific radiative transfer mode (RTTOV) coefficient file is matched for it.

[0101] The initial or forecast field is generated using the regional model WRF and used as the background field in the assimilation process.

[0102] We collected and organized real-time land surface temperature products retrieved from the AGRI thermal infrared channel of Fengyun-4B satellite to correct the land surface temperature of the subsequent background field.

[0103] The land surface temperature of the wind-driven four-phase surface was interpolated to the spatial grid points of the background field using a bilinear interpolation method, and the spatiotemporal resolution of the two was matched.

[0104] The 2-meter temperature and humidity, surface radiation, surface heat flux and wind speed, and land surface temperature, which are closely related to land surface temperature in the model, will be used as meteorological feature vectors for subsequent correction of land surface temperature.

[0105] Historical wind-induced land surface temperatures and WRF-simulated meteorological features for corresponding time periods were used as the training set. A random forest algorithm was employed to fit the data to the training samples, and the land surface temperature magnitude corresponding to the meteorological feature vector was used as the prediction result to obtain a machine learning correction model for land surface temperature.

[0106] Due to factors such as clouds and precipitation, the Feng-4 land surface temperature product often has a large number of missing values. The missing values ​​are determined by the spatiotemporally matched Feng-4 land surface temperature.

[0107] In the non-missing region of the wind-induced land surface temperature, its value replaces the land surface temperature of the background field at the corresponding time.

[0108] In the missing land surface temperature region of Wind 4, the land surface temperature of the background field is corrected using a step-by-step land surface temperature correction model.

[0109] By combining and The optimization of the background field land surface temperature was completed.

[0110] Based on historical background field data of the study area, the background error calculation model in WRFDA is used to construct a background error covariance matrix describing the errors of various variables (such as temperature, humidity, wind field, etc.) in the background field and their spatial correlation, and the background error is calculated to obtain background error information.

[0111] Steps , and The obtained coefficient file, the optimized background field, and the background error are input into the observation operator (RTTOV). The observation operator converts the background field variables (such as temperature, humidity, air pressure, etc.) into simulated radiance and further calculates the difference between the observed and the simulated radiance of the background field to obtain the observation increment (OB). Excessively large observation increments are removed through quality control.

[0112] The objective function is calculated by combining the observation data with the background field using a three-dimensional variational method.

[0113] Through multiple iterations, the objective function is minimized after convergence, generating an optimal analysis field that closely approximates the atmospheric "true value".

[0114] By outputting analysis fields and diagnostic files through WRFDA, the error characteristics of observation increments and analysis increments (the difference between the emissivity of the observation and analysis fields) are statistically analyzed, the degree of improvement of variables such as atmospheric temperature in the analysis field is evaluated, and the direct assimilation of the Fengyun spaceborne microwave hygrometer is completed.

[0115] Based on any of the above embodiments, the background field land surface temperature correction process includes: replacing the background field land surface temperature of the corresponding grid points with the spatiotemporally matched Fengyun-4B land surface temperature; for the missing Fengyun-4B land surface temperature areas, the background field land surface temperature is corrected using machine learning methods, and its root mean square error is significantly reduced after correction. Figure 3 As shown, the land surface temperature of Fengyun-4B (top left), the background field land surface temperature (top right), the corrected background field land surface temperature (bottom left), and the root mean square error of land surface temperature after machine learning correction (bottom right).

[0116] After optimizing the background field land surface temperature, the assimilation system significantly improved the quantity and quality of direct assimilation of data from the Fengyun spacecraft-borne microwave hygrometer window area and near-surface channel. For example... Figure 4 As shown, the assimilation observation points before optimization (first row), the assimilation observation points after optimization (second row), and the optimized increment (third row) are displayed. Different columns represent different microwave radiometer channels. Figure 5As shown, the assimilation quantity (left) and root mean square error (RMS) before and after optimization are shown. After optimizing the background field land surface temperature, the quality of the analyzed field atmospheric temperature is significantly improved. The vertical distribution of the RMS error of the analyzed field atmospheric temperature profile before and after optimization is shown in the figure. Figure 6 As shown.

[0117] The Fengyun satellite observation data assimilation method provided in this invention establishes a correction module for land surface temperature in a fast radiative transfer mode, improving the quantity and quality of direct assimilation from Fengyun satellite-borne microwave hygrometers. These results will further promote the application of Fengyun satellite data in weather forecasting for the plateau and its downstream areas, enhancing my country's meteorological disaster forecasting capabilities. Simultaneously, it can also be used for numerical simulation and analysis of meteorological disasters in complex terrain, which is of great significance for the development of relevant reanalysis data. Furthermore, the improved observation operator approach in this invention is universally applicable, and in the future, this method can be extended to the direct assimilation of more satellite series and satellite-borne microwave instruments, further increasing the practical application rate of satellite data. This invention addresses the technical challenge of assimilating surface-sensitive spaceborne microwave observations in complex terrain. It innovatively utilizes the land surface temperature products from the new-generation Fengyun-4B satellite and machine learning methods to develop a highly timely background field land surface temperature correction algorithm. Furthermore, it integrates a corresponding observation operator optimization module within the existing mainstream data assimilation system (WRFDA). This enables real-time assimilation of Fengyun microwave hygrometer data during numerical weather prediction for different underlying surfaces or regions, significantly improving the application rate and assimilation quality of Fengyun-3 polar-orbiting satellite microwave hygrometer data. The method is convenient and efficient, and can be subsequently applied to the direct assimilation of different spaceborne microwave hygrometers, making it possible to extract large amounts of land information from spaceborne microwave observations.

[0118] The Fengyun satellite observation data assimilation device provided by the present invention is described below. The Fengyun satellite observation data assimilation device described below and the Fengyun satellite observation data assimilation method described above can be referred to in correspondence with each other.

[0119] Figure 7 This is a schematic diagram of the structure of the Fengyun satellite observation data assimilation device provided in an embodiment of the present invention, as shown below. Figure 7 As shown, the Fengyun satellite observation data assimilation device provided in this embodiment of the invention includes:

[0120] The generation module 701 is used to generate a numerical model background field using a region model.

[0121] Alignment module 702 is used to obtain the real-time land surface temperature from the thermal infrared channel of the first satellite observation equipment, and to align the real-time land surface temperature with the spatial grid of the background field of the numerical model in terms of spatiotemporal resolution to obtain the measured area of ​​land surface temperature that matches the spatial grid of the background field.

[0122] The optimization module 703 is used to replace the land surface temperature in the background field at the corresponding time with the real-time land surface temperature corresponding to the non-missing measurement area if the measured area is a non-missing measurement area; and to correct the land surface temperature in the background field at the corresponding time based on the land surface temperature correction model if the measured area is a non-missing measurement area, so as to obtain an optimized background field.

[0123] The construction module 704 is used to construct errors describing each variable in the background field based on historical background field data, obtain background error information, and convert the optimized background field and the background error information into simulated emissivity.

[0124] The acquisition module 705 is used to collect raw radiance observation data acquired by the second satellite observation equipment, and to preprocess the raw radiance observation data to obtain the observed radiance.

[0125] The output module 706 is used to combine the observed radiance with the optimized background field to obtain an analysis field, iteratively optimize the analysis field, and output the optimal analysis field.

[0126] The land surface temperature correction model is trained based on historical land surface temperatures derived from inversion and meteorological features simulated by regional models for the corresponding time periods.

[0127] The Fengyun satellite observation data assimilation device provided in this embodiment of the invention generates a numerical model background field using a regional model; obtains real-time land surface temperature from the thermal infrared channel of a first satellite observation device; aligns the real-time land surface temperature with the spatial grid of the numerical model background field using spatiotemporal resolution to obtain a measured area of ​​land surface temperature matching the spatial grid of the background field; if the measured area is a non-missing area, the real-time land surface temperature corresponding to the non-missing area is used to replace the land surface temperature in the background field at the corresponding time; if the measured area is a missing area, the meteorological element characteristics in the background field at the corresponding time are corrected based on a land surface temperature correction model to obtain an optimized background field; and constructs an error description of each variable in the background field based on historical background field data to obtain background error information. The optimized background field and background error information are converted into simulated radiance. Raw radiance observation data collected by a second satellite observation device are collected and preprocessed to obtain observed radiance. The observed radiance is combined with the optimized background field to obtain an analysis field. The analysis field is iteratively optimized to output the optimal analysis field. The land surface temperature correction model is trained based on historically inverted land surface temperatures and meteorological element characteristics simulated by regional models for the corresponding time period. This invention reduces background field errors by optimizing the land surface temperature parameters, allowing more satellite microwave observation data to enter the assimilation system, thus improving the system's satellite data assimilation capabilities and enhancing the assimilation effect of satellite data under complex high-altitude terrain.

[0128] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840. The processor 810, communication interface 820, and memory 830 communicate with each other via the communication bus 840. The memory 830 includes computer programs, an operating system, and acquired data. The processor 810 can call logical instructions in the memory 830 to execute a Fengyun satellite observation data assimilation method. This method includes: generating a numerical model background field using a regional model; obtaining real-time land surface temperature from the thermal infrared channel of a first satellite observation device; aligning the real-time land surface temperature with the spatial grid points of the numerical model background field using spatiotemporal resolution to obtain a measured area of ​​land surface temperature matching the spatial grid points of the background field; if the measured area is a non-missing area, replacing the land surface temperature in the background field at the corresponding time with the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, then using a land surface temperature correction model... The land surface temperature at the corresponding time in the background field is corrected to obtain an optimized background field; errors describing each variable in the background field are constructed based on historical background field data to obtain background error information; the optimized background field and the background error information are converted into simulated radiance; raw radiance observation data collected by a second satellite observation device are collected, and the raw radiance observation data are preprocessed to obtain observed radiance; the observed radiance is combined with the optimized background field to obtain an analysis field; the analysis field is iteratively optimized to output the optimal analysis field; wherein, the land surface temperature correction model is trained based on historically inverted land surface temperatures and meteorological element characteristics simulated by regional models for the corresponding time period.

[0129] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to related technologies, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0130] On the other hand, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the Fengyun satellite observation data assimilation method provided by the above methods. The method includes: generating a numerical model background field using a regional model; obtaining the real-time land surface temperature from the thermal infrared channel of a first satellite observation device; aligning the real-time land surface temperature with the spatial grid of the numerical model background field using spatiotemporal resolution to obtain a measured area of ​​land surface temperature matching the spatial grid of the background field; if the measured area is a non-missing area, replacing the land surface temperature in the background field at the corresponding time with the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, then based on the land surface temperature... The temperature correction model corrects the land surface temperature at the corresponding time in the background field to obtain an optimized background field; based on historical background field data, it constructs an error description of each variable in the background field to obtain background error information, and converts the optimized background field and the background error information into simulated radiance; it collects raw radiance observation data acquired by a second satellite observation device, preprocesses the raw radiance observation data to obtain observed radiance; it combines the observed radiance with the optimized background field to obtain an analysis field, iteratively optimizes the analysis field, and outputs the optimal analysis field; wherein, the land surface temperature correction model is trained based on historically inverted land surface temperatures and meteorological element characteristics simulated by regional models for the corresponding time period.

[0131] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0132] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of software products. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assimilating Fengyun satellite observation data, characterized in that, include: Generate the background field of the numerical model using the region model; The real-time land surface temperature is retrieved from the thermal infrared channel of the first satellite observation equipment. The real-time land surface temperature is then aligned with the spatial grid of the background field of the numerical model to obtain the measured area of ​​land surface temperature that matches the spatial grid of the background field. If the measured area is a non-missing area, the real-time land surface temperature corresponding to the non-missing area is used to replace the land surface temperature in the background field at the corresponding time; if the measured area is a missing area, the land surface temperature in the background field at the corresponding time is corrected based on the land surface temperature correction model to obtain an optimized background field. Based on historical background field data, errors describing the variables in the background field are constructed to obtain background error information. The optimized background field and the background error information are then converted into simulated radiance. Specifically, this includes: matching the coefficient file of the radiative transfer mode according to the type of the second satellite observation equipment; inputting the coefficient file, the optimized background field, and the background error information into the observation operator, and converting the background field variables into simulated radiance through the observation operator. Collect raw radiance observation data acquired by the second satellite observation equipment, preprocess the raw radiance observation data, and obtain the observed radiance; The observed radiance is combined with the optimized background field to obtain the analysis field. The analysis field is then iteratively optimized to output the optimal analysis field. The land surface temperature correction model is trained based on historical land surface temperatures derived from inversion and meteorological features simulated by regional models for the corresponding time periods.

2. The Fengyun satellite observation data assimilation method according to claim 1, characterized in that, The correction of the land surface temperature in the background field at the corresponding time based on the land surface temperature correction model includes: Extract feature vectors of meteorological elements closely related to land surface temperature from numerical models; Collect land surface temperature data retrieved from historical time periods, as well as meteorological feature data simulated by regional models for the corresponding time periods, and construct a training set; The random forest algorithm was used to fit the training samples in the training set to obtain the land surface temperature correction model. The meteorological element feature vectors are input into the land surface temperature correction model to obtain the predicted land surface temperature, which is then used as the optimized land surface temperature.

3. The Fengyun satellite observation data assimilation method according to claim 1, characterized in that, The preprocessing of the original radiance observation data to obtain the observed radiance includes: The raw radiance observation data is filtered; The observed radiance is obtained by correcting the bias in the filtered radiance observation data using the dynamic bias correction method.

4. The Fengyun satellite observation data assimilation method according to claim 1, characterized in that, The process of constructing an error description of each variable in the background field based on historical background field data, thereby obtaining background error information, includes: Based on historical background field data of the study area, key variables in the pattern background field are extracted; Using the background error calculation module in the WRFDA assimilation system, the error values ​​and spatial correlations of the key variables are calculated. Based on the error values ​​and spatial correlation, a background error covariance matrix is ​​constructed. The covariance matrix is ​​used to describe the error distribution of each variable in the mode background field and their interrelationships. The background error covariance matrix is ​​standardized using the WRFDA assimilation system to generate background error information.

5. The Fengyun satellite observation data assimilation method according to claim 1, characterized in that, After obtaining the observed emissivity, the following is also included: The difference between the observed radiance and the simulated radiance is calculated to obtain the observation increment; Observation points with observation increments greater than a preset threshold on the spatial grid are removed, and the observation radiance used for subsequent assimilation is selected. The selected observed radiance is combined with the optimized background field to obtain the analysis field.

6. The Fengyun satellite observation data assimilation method according to claim 5, characterized in that, The iterative optimization of the analysis field to output the optimal analysis field includes: The selected observed radiance is combined with the optimized background field using a three-dimensional variational method to construct an objective function, and the analysis field is calculated based on the objective function. The optimized background field is replaced by the analytical field, and then recombined with the screened observed radiance. The objective function is iteratively calculated to minimize the objective function, thereby obtaining the optimal analytical field.

7. A device for assimilating Fengyun satellite observation data, characterized in that, include: The generation module is used to generate the background field of the numerical model using the regional model; The alignment module is used to obtain the real-time land surface temperature from the thermal infrared channel of the first satellite observation equipment, and to align the real-time land surface temperature with the spatial grid of the background field of the numerical model in terms of spatiotemporal resolution, so as to obtain the measured area of ​​land surface temperature that matches the spatial grid of the background field. The optimization module is used to replace the land surface temperature in the background field at the corresponding time with the real-time land surface temperature corresponding to the non-missing measurement area if the measured area is a non-missing measurement area; and to correct the land surface temperature in the background field at the corresponding time based on the land surface temperature correction model if the measured area is a non-missing measurement area, so as to obtain an optimized background field. The construction module is used to construct errors describing each variable in the background field based on historical background field data, obtain background error information, and convert the optimized background field and the background error information into simulated radiance. Specifically, it includes: matching the coefficient file of the radiative transfer mode according to the type of the second satellite observation equipment; inputting the coefficient file, the optimized background field, and the background error information into the observation operator, and converting the background field variables into simulated radiance through the observation operator; The acquisition module is used to collect raw radiance observation data acquired by the second satellite observation equipment, and to preprocess the raw radiance observation data to obtain the observed radiance. The output module is used to combine the observed radiance with the optimized background field to obtain the analysis field, iteratively optimize the analysis field, and output the optimal analysis field. The land surface temperature correction model is trained based on historical land surface temperatures derived from inversion and meteorological features simulated by regional models for the corresponding time periods.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the Fengyun satellite observation data assimilation method as described in any one of claims 1 to 6.

9. A non-transitory readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the Fengyun satellite observation data assimilation method as described in any one of claims 1 to 6.

Images