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

Abstract

The invention provides a wind cloud satellite observation data assimilation method, a wind cloud satellite observation data assimilation device, electronic equipment and a storage medium, and relates to the technical field of meteorological data processing. Real-time land surface temperature and space lattice points of a background field are subjected to temporal-spatial resolution alignment; if so, replacing the land surface temperature in the background field at the corresponding moment with the real-time land surface temperature corresponding to the non-missing measurement area; if the actual measurement area is a missing measurement area, correcting the land surface temperature at the corresponding moment in the background field based on a land surface temperature correction model to obtain an optimized background field; combining the observed radiance with the optimized background field, constructing an objective function, and calculating to obtain an analysis field; and carrying out iterative optimization on the analysis field to minimize the target function, and outputting an optimal analysis field. According to the method, background field errors can be reduced, so that the system has better satellite data assimilation capability, and meanwhile, the assimilation effect of the satellite data under the high-altitude complex terrain is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological data processing, and in particular to a method, device, electronic equipment and storage medium for assimilating Fengyun satellite observation data. Background Art

[0002] Weather forecasting has important social and economic benefits in the fields of agriculture, transportation, electricity, etc. At present, the main method of weather forecasting is numerical forecasting, which describes the dynamic and thermal processes of the atmosphere through a set of equations, and uses computers to solve them under certain initial and boundary conditions to obtain the forecast values ​​of various elements in the atmosphere within a certain period of time in the future. Generally speaking, for medium- and short-term time scales with a forecast duration of less than one week, numerical forecasting belongs to the initial value problem of differential equations, and the accuracy of the initial field has a decisive influence on the forecast effect. Assimilation technology is the process of integrating multiple observation data (such as satellite observations, ground station observations) with numerical models to produce a numerical model of the atmospheric state that is closer to the actual situation, providing a more accurate initial field for forecasting. In order to reduce the initial field error, it is necessary to put as many observation data as possible into the assimilation system.

[0003] In related technologies, the atmospheric radiation transfer equation is parameterized to construct the relationship between atmospheric state variables and satellite observed radiance, and this relationship is put into the assimilation system as an observation operator to directly assimilate the satellite observed radiance data. As observation operators of satellite-borne microwave radiometers, fast radiation 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). When the atmospheric state variables in the assimilation background field are given, it can be used to quickly calculate the satellite radiation brightness temperature. However, due to its high altitude and complex terrain, the simulation accuracy of land surface temperature and land surface emissivity in the radiation transfer model is often low in the plateau area, while they are important parameters for calculating the atmospheric emissivity. This makes it impossible for a large part of the satellite observations in some areas, such as the plateau, to enter the assimilation system, resulting in the lack of satellite observation information in numerical forecasts. The errors caused by the radiation transfer model calculation are large, and the assimilation effect of satellite data in high-altitude complex terrain is poor. Summary of the invention

[0004] The present invention provides a Fengyun satellite observation data assimilation method, device, electronic device and storage medium, which are used to solve the defects of the traditional Fengyun satellite observation data assimilation method, such as the lack of satellite observation information, large calculation error of radiation transmission mode, and poor assimilation effect of satellite data under high altitude and complex terrain.

[0005] The present invention provides a method for assimilating Fengyun satellite observation data, comprising: Generate the background field of numerical model using regional model; Acquire the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, align the real-time land surface temperature with the spatial grid points of the numerical model background field in time and space resolution, and obtain the measured area of ​​the land surface temperature matching the spatial grid points of the background field; If the measured area is a non-missing area, the land surface temperature in the background field at the corresponding time is replaced by the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, the land surface temperature at the corresponding time in the background field is corrected based on the land surface temperature correction model to obtain an optimized background field; Based on historical background field data, an error describing each variable in the background field is constructed to obtain background error information, and the optimized background field and the background error information are converted into simulated radiance; Collecting original radiation rate observation data collected by the second satellite observation device, and preprocessing the original radiation rate observation data to obtain observed radiation rate; Calculating the difference between the observed radiance and the simulated radiance to obtain an observed increment; Combining the observed radiance with the optimized background field to obtain an analysis field, iteratively optimizing the analysis field, and outputting an optimal analysis field; The land surface temperature correction model is trained based on the land surface temperature inverted from history and the meteorological element characteristics simulated by the regional model of the corresponding time period.

[0006] According to the Fengyun satellite observation data assimilation method provided by the present invention, the land surface temperature at the corresponding time in the background field is corrected based on the land surface temperature correction model, including: Extract the characteristic vectors of meteorological elements closely related to land surface temperature from numerical models; Collect the inverted land surface temperature data in the historical time period and the characteristic data of meteorological elements simulated by the regional model in the corresponding time period to construct a training set; The random forest algorithm is used to fit the training samples in the training set to obtain a land surface temperature correction model; The meteorological element characteristic vector is input into the land surface temperature correction model to obtain the predicted land surface temperature as the optimized land surface temperature.

[0007] According to the Fengyun satellite observation data assimilation method provided by the present invention, the raw radiance observation data is preprocessed to obtain the observed radiance, including: filtering the original radiance observation data; The observed radiance data after filtering is corrected using the dynamic deviation correction method to obtain the observed radiance.

[0008] According to the Fengyun satellite observation data assimilation method provided by the present invention, the error describing each variable in the background field is constructed based on the historical background field data to obtain background error information, including: Based on the historical background field data of the study area, key variables in the model background field are extracted; Using the background error calculation module in the WRFDA assimilation system, the error values ​​of the key variables and their spatial correlations are calculated; Based on the error values ​​and spatial correlation, a background error covariance matrix is ​​constructed, wherein the covariance matrix is ​​used to describe the error distribution of each variable in the model background field and their mutual relationship; The background error covariance matrix is ​​standardized by the WRFDA assimilation system to generate background error information.

[0009] 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 radiance includes: According to the type of the second satellite observation device, matching the coefficient file of the radiation transfer model; The coefficient file, the optimized background field and the background error information are input into an observation operator, and the background field variables are converted into simulated radiance through the observation operator.

[0010] According to the Fengyun satellite observation data assimilation method provided by the present invention, after obtaining the observed radiance, Calculate the difference between the observed radiance and the simulated radiance to obtain an observed increment, Eliminate observation points on the spatial grid whose observation increment is greater than the preset threshold, and select the observed radiances for subsequent assimilation; The screened observed radiance is combined with the optimized background field to obtain the analysis field.

[0011] According to the Fengyun satellite observation data assimilation method provided by the present invention, the iterative optimization of the analysis field and output of the optimal analysis field comprises: Combining the screened observed radiance with the optimized background field using a three-dimensional variational method, constructing an objective function, and calculating an analysis field based on the objective function; The optimized background field is replaced by the analysis field, and the field is recombined with the screened observed radiance. The objective function is iteratively calculated to minimize the objective function, thereby obtaining the optimal analysis field.

[0012] The present invention also provides a Fengyun satellite observation data assimilation device, comprising: A generation module, used for generating a numerical model background field using a regional model; An alignment module is used to obtain the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, align the real-time land surface temperature with the spatial grid points of the numerical model background field in time and space resolution, and obtain a measured area of ​​the land surface temperature matching the spatial grid points of the background field; an optimization module, for 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 non-missing area; and for correcting the land surface temperature at the corresponding time in the background field based on the land surface temperature correction model if the measured area is a missing area, so as to obtain an optimized background field; A construction module, used to construct an error 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; A collection module, used to collect the original radiation rate observation data collected by the second satellite observation device, and pre-process the original radiation rate observation data to obtain the observed radiation rate; An output module, used for combining the observed radiance with the optimized background field to obtain an analysis field, iteratively optimizing the analysis field, and outputting an optimal analysis field; The land surface temperature correction model is trained based on the land surface temperature inverted from history and the meteorological element characteristics simulated by the regional model of the corresponding time period.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the Fengyun satellite observation data assimilation method as described in any one of the above items is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for assimilating Fengyun satellite observation data described in any one of the above items is implemented.

[0015] The Fengyun satellite observation data assimilation method, device, electronic device and storage medium provided by the present invention generate a numerical model background field by using a regional model; obtain the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, align the time and space resolution of the real-time land surface temperature with the spatial grid points of the numerical model background field, and 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, use the real-time land surface temperature corresponding to the non-missing area to replace the land surface temperature in the background field at the corresponding time; if the measured area is a missing area, correct the land surface temperature at the corresponding time in the background field based on the land surface temperature correction model to obtain an optimized background field; construct an error description of each variable in the background field based on historical background field data to obtain a background field. error information, converting the optimized background field and the background error information into simulated emissivity; collecting original emissivity observation data collected by the second satellite observation equipment, preprocessing the original emissivity observation data to obtain observed emissivity; combining the observed emissivity with the optimized background field to obtain an analysis field, iteratively optimizing the analysis field, and outputting an optimal analysis field; wherein the land surface temperature correction model is trained based on the land surface temperature inverted from history and the meteorological element characteristics simulated by the regional model of the corresponding time period. The present 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, so that the system has better satellite data assimilation capabilities, and at the same time improves the assimilation effect of satellite data under high-altitude complex terrain. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is one of the flow charts of the Fengyun satellite observation data assimilation method provided by an embodiment of the present invention; Figure 2 This is the second flow chart of the Fengyun satellite observation data assimilation method provided by the embodiment of the present invention; Figure 3 is a schematic diagram of background field land surface temperature correction provided by an embodiment of the present invention; Figure 4 It is a schematic diagram of direct assimilation of the window area and near-surface channel data of the Fengyun satellite-borne microwave hygrometer provided by an embodiment of the present invention; Figure 5 Schematic diagram of assimilation quantity and root mean square error before and after optimization provided by an embodiment of the present invention; Figure 6 is a schematic diagram of vertical distribution of the root mean square error of the atmospheric temperature profile of the analysis field before and after optimization provided by an embodiment of the present invention; Figure 7 Schematic diagram of the functional structure of a Fengyun satellite observation data assimilation device provided by an embodiment of the present invention; Figure 8 It is a functional structure diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Figure 1 The flowchart of the Fengyun satellite observation data assimilation method provided by the embodiment of the present invention is as follows: Figure 1 As shown, the Fengyun satellite observation data assimilation method provided by the embodiment of the present invention includes: Step 101, generating a numerical model background field using a regional model; Step 102: obtaining the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, aligning the real-time land surface temperature with the spatial grid points of the numerical model background field in time and space resolution, and obtaining the measured area of ​​the land surface temperature matching the spatial grid points of the background field; Due to the influence of factors such as clouds and precipitation, there are often a large number of missing values ​​in the Feng Si Lu surface temperature products. The missing values ​​of the Feng Si Lu surface temperature after time and space matching are judged.

[0020] Step 103: if the measured area is a non-missing area, the land surface temperature in the background field at the corresponding time is replaced by the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, the land surface temperature at the corresponding time in the background field is corrected based on the land surface temperature correction model to obtain an optimized background field; Step 104: construct an error describing 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 radiance; Step 105: collecting the original radiation rate observation data collected by the second satellite observation device, and preprocessing the original radiation rate observation data to obtain the observed radiation rate; Step 106: combining the observed radiance with the optimized background field to obtain an analysis field, iteratively optimizing the analysis field, and outputting an optimal analysis field; The land surface temperature correction model is trained based on the land surface temperature inverted from history and the meteorological element characteristics simulated by the regional model of the corresponding time period.

[0021] Traditional assimilation technology parameterizes the atmospheric radiation transfer equation, constructs the relationship between atmospheric state variables and satellite observed radiance, and puts this relationship into the assimilation system as an observation operator to directly assimilate satellite observed radiance data. As the observation operator of the satellite-borne microwave radiometer, the fast radiation transfer mode such as RTTOV and CRTM is widely used in advanced assimilation systems such as WRFDA and GSI. When the atmospheric state variables in the assimilation background field are given, it can be used to quickly calculate the satellite radiation brightness temperature. However, due to its high altitude and complex terrain, the simulation accuracy of land surface temperature and land surface emissivity in the radiation transfer model is often low in the plateau area, and they are important parameters for calculating atmospheric radiance. This makes some areas such as a large part of the satellite observations over the plateau fail to enter the assimilation system, resulting in the lack of satellite observation information in the numerical forecast. The error caused by the radiation transfer model calculation is large, and the assimilation effect of satellite data in high altitude and complex terrain is poor.

[0022] The Fengyun satellite observation data assimilation method provided in an embodiment of the present invention generates a numerical model background field by using a regional model; obtains the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, aligns the time and space resolution of the real-time land surface temperature with the spatial grid points of the numerical model background field, and obtains 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, 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 moment; if the measured area is a missing area, the land surface temperature at the corresponding moment in the background field is corrected based on a land surface temperature correction model to obtain an optimized background field; based on historical background field data, an error describing each variable in the background field is constructed to obtain background error information, The optimized background field and the background error information are converted into simulated radiance; the original radiance observation data collected by the second satellite observation equipment are collected, and the original radiance observation data are preprocessed to obtain the observed radiance; the observed radiance is combined with the optimized background field to obtain the 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 the land surface temperature obtained by historical inversion and the characteristics of meteorological elements simulated by the regional model of the corresponding time period. The present invention reduces the error of the background field by optimizing the land surface temperature parameters of the background field, allows more satellite microwave observation data to enter the assimilation system, enables the system to have better satellite data assimilation capabilities, and at the same time improves the assimilation effect of satellite data under high-altitude complex terrain.

[0023] Based on any of the above embodiments, the correcting the land surface temperature at the corresponding time in the background field based on the land surface temperature correction model includes: Step 201: extracting meteorological characteristic vectors closely related to land surface temperature from the numerical model; Step 202: collect the inverted land surface temperature data in the historical time period and the characteristic data of the regional model simulated meteorological elements in the corresponding time period to construct a training set; Step 203: using a random forest algorithm to fit the training samples in the training set to obtain a land surface temperature correction model; Step 204: input the meteorological element characteristic vector into the land surface temperature correction model to obtain the predicted land surface temperature as the optimized land surface temperature.

[0024] In the embodiment of the present invention, the 2-meter temperature and humidity, surface radiation, surface heat flux, wind speed and land surface temperature, which are closely related to the land surface temperature in the forecast model, are used as the meteorological element feature vectors for the subsequent correction of the land surface temperature. The historical wind speed and land surface temperature and the WRF simulated meteorological element feature data of the corresponding time period are used as the training set, and the random forest algorithm is used to fit the training samples, and the land surface temperature corresponding to the feature vector is used as the prediction result to obtain the machine learning correction model of the land surface temperature.

[0025] The embodiment of the present invention combines the recent wind-four land surface temperature products with high spatiotemporal resolution and high timeliness with machine learning methods to achieve real-time generation of seamless land surface temperature optimization products with high spatiotemporal resolution. In the past, there was a lack of real-time, spatially seamless land surface temperature observations in many land areas. Therefore, in the implementation of business or research on direct assimilation of satellite-borne microwave observations, it was difficult to correct the background field land surface temperature in a timely manner, wasting a large amount of satellite-borne microwave observation resources. The embodiment of the present invention combines the recent wind-four land surface temperature products with high spatiotemporal resolution and high timeliness with machine learning methods to achieve real-time generation of seamless land surface temperature optimization products, directly optimize and correct the background field land surface temperature, and enable more satellite-borne microwave observation data to be assimilated.

[0026] Based on any of the above embodiments, the preprocessing of the original radiation rate observation data to obtain the observed radiation rate includes: Step 301, filtering the original radiance observation data; Step 302: Perform deviation correction on the radiometric observation data filtered by the dynamic deviation correction method to obtain the observed radiometric data.

[0027] The embodiment of the present invention inputs the radiance observation data into the data preprocessing module for decoding and reading, extracts its observation information, performs quality control, eliminates obviously invalid data and excessive data of the satellite observation zenith angle, and then uses the dynamic deviation correction method (VARBC) to correct the deviation of the observation data, thereby obtaining accurate observed radiance.

[0028] Based on any of the above embodiments, constructing an error describing each variable in the background field based on historical background field data to obtain background error information includes: Step 401: extract key variables in the model background field based on the historical background field data of the study area; Step 402: Calculate the error value and spatial correlation of the key variable using the background error calculation module in the WRFDA assimilation system; Step 403: constructing a background error covariance matrix based on the error values ​​and spatial correlation, wherein the covariance matrix is ​​used to describe the error distribution of each variable in the pattern background field and their mutual relationship; Step 404: Standardize the background error covariance matrix through the WRFDA assimilation system to generate background error information.

[0029] Based on any of the above embodiments, converting the optimized background field and the background error information into simulated radiance includes: Step 501: Match the coefficient file of the radiation transmission mode according to the type of the second satellite observation device; Step 502: input the coefficient file, the optimized background field and the background error information into an observation operator, and convert the background field variables into simulated radiance through the observation operator.

[0030] The embodiment of the present invention constructs a background field land surface temperature correction module in the mainstream data assimilation system (WRFDA) and the rapid radiative transfer model (RTTOV). The background field land surface temperature output by the model enters the observation operator after optimization, so that the new system can assimilate more window area and near-surface channel data of the FY satellite-borne microwave hygrometer, which can promote the application of land surface information in weather forecasting.

[0031] Traditionally, the deviation of the land surface temperature of the background field is adjusted indirectly by improving the parameterization scheme of the physical processes related to the land surface temperature in the model. However, the optimal parameterization scheme for the land surface temperature may be different in different regions, requiring a lot of testing and difficult model development work. In addition, changes in the parameterization scheme will also cause changes in other variables in the model, complicating the problem. In the embodiment of the present invention, the relevant modules for optimizing the land surface temperature of the background field are independent of other modules, and the optimization and correction process will not affect other variables in the model. In addition, the optimization method is applicable to different regions and has universal applicability. In addition, the optimization and correction scheme in the embodiment of the present invention is mainly based on machine learning, which can reduce a lot of time costs and computing resources compared with previous methods such as testing different parameterization schemes.

[0032] Based on any of the above embodiments, after obtaining the observed emissivity, the method further includes: Step 601, calculating the difference between the observed radiance and the simulated radiance to obtain an observed increment; Step 602: remove observation points with too large observation increments on the spatial grid, select the observed radiances for subsequent assimilation, and combine them with the background field after optimizing the land surface temperature to obtain the analysis field.

[0033] The embodiment of the present invention can effectively reduce the observation increment by optimizing the land surface temperature parameter in the observation operator, and at the same time improve the background field quality, allow more satellite radiance observation data to enter the assimilation system.

[0034] Based on any of the above embodiments, the iterative optimization of the analysis field to output the optimal analysis field includes: Step 701: Combining the screened observed emissivity with the background field after optimizing the land surface temperature using a three-dimensional variational method, constructing an objective function to calculate an analysis field; Step 702: Replace the background field after optimizing the land surface temperature with the analysis field, re-combine with the screened observed radiance, iteratively calculate the objective function, minimize the objective function, and obtain the optimal analysis field.

[0035] Based on any of the above embodiments, Figure 2 As shown, the Fengyun satellite observation data assimilation method provided by the embodiment of the present invention specifically includes: The original emissivity observation data of the FY-3 microwave hygrometer were collected and collated for subsequent satellite data assimilation.

[0036] The radiance observation data is input into the data preprocessing module for decoding and reading, and its observation information is extracted for quality control. Obviously invalid data and excessive data of satellite observation zenith angle are eliminated, and the dynamic deviation correction method (VARBC) is used to correct the deviation of the observation data.

[0037] According to the type of spaceborne microwave hygrometer instrument, a specific radiation transfer mode (RTTOV) coefficient file is matched for it.

[0038] The regional model WRF is used to generate the initial or forecast fields, which are used as the background fields in the assimilation process.

[0039] The real-time land surface temperature products inverted by the AGRI thermal infrared channel of FY-4B are collected and collated for correcting the land surface temperature of subsequent background fields.

[0040] The bilinear interpolation method is used to interpolate the wind surface temperature to the spatial grid of the background field to match the temporal and spatial resolutions of the two.

[0041] The 2-meter temperature and humidity, surface radiation, surface heat flux wind speed, and land surface temperature in the model, which are closely related to the land surface temperature, are used as the characteristic vectors of meteorological elements for subsequent correction of the land surface temperature.

[0042] The historical wind power land surface temperatures and the WRF simulated meteorological element characteristics of the corresponding time period were taken as training sets. The random forest algorithm was used to fit the training samples, and the land surface temperature corresponding to the meteorological element feature vector was used as the prediction result to obtain a machine learning correction model of the land surface temperature.

[0043] Due to the influence of factors such as clouds and precipitation, there are often a large number of missing values ​​in the Feng Si Lu surface temperature products. The missing values ​​of the Feng Si Lu surface temperature after time and space matching are judged.

[0044] In areas where the land surface temperature of Fengsi is not missing, its value is used to replace the land surface temperature of the background field at the corresponding time.

[0045] In the areas where the land surface temperature of Fengsi is missing, the land surface temperature correction model is used to correct the land surface temperature of the background field.

[0046] By combining and , complete the optimization of the background field land surface temperature.

[0047] According to the 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 and spatial correlations of various variables in the model background field (such as temperature, humidity, wind field, etc.), complete the background error calculation, and obtain the background error information.

[0048] Step , and The obtained coefficient file, optimized background field and 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 background field simulated radiances to obtain the observation increment (OB). Excessive observation increments are eliminated through quality control.

[0049] The observed data are combined with the background field using a three-dimensional variational method to calculate the objective function.

[0050] Through multiple iterations, the objective function is minimized after convergence, generating an optimal analysis field close to the "true value" of the atmosphere.

[0051] Through the output of analysis fields and diagnostic files by WRFDA, the error characteristics of observation increment and analysis increment (the difference between the observed and analysis field radiance) are statistically analyzed, the improvement degree of variables such as atmospheric temperature in the analysis field is evaluated, and the direct assimilation of the FY satellite-borne microwave hygrometer is completed.

[0052] Based on any of the above embodiments, the background field land surface temperature correction process includes: using the FY-4B land surface temperature after time and space matching to replace the background field land surface temperature of the corresponding grid point, and for the area where the FY-4B land surface temperature is missing, the background field land surface temperature is corrected by machine learning method, and its root mean square error is significantly reduced after correction. Figure 3 As shown, the FY-4B land surface temperature (upper left), the background field land surface temperature (upper right), the corrected background field land surface temperature (lower left), and the root mean square error of the land surface temperature after machine learning correction (lower right).

[0053] After optimizing the background field land surface temperature, the assimilation system has significantly improved the quantity and quality of direct assimilation of the FY satellite-borne microwave hygrometer window area and near-surface channel data. Figure 4 As shown in the figure, the assimilated observation points before optimization (first row), the assimilated observation points after optimization (second row), and the increment after optimization (third row). Different columns represent different microwave radiometer channels. Figure 5 As shown in the figure, the assimilation quantity (left) and root mean square error (right) before and after optimization. After optimizing the background field land surface temperature, the quality of the analysis field atmospheric temperature is significantly improved. The vertical distribution of the root mean square error of the analysis field atmospheric temperature profile before and after optimization is shown in the figure. Figure 6 shown.

[0054] The Fengyun satellite observation data assimilation method provided in the embodiment of the present invention has established a correction module for the land surface temperature in the rapid radiation transfer mode, and has achieved an improvement in the quantity and quality of the direct assimilation of the Fengyun satellite-borne microwave hygrometer. The relevant achievements will further promote the application of Fengyun satellite data in weather forecasts in the plateau and its downstream, and improve the forecast level of meteorological disasters in my country. At the same time, it can also be used for numerical simulation and analysis of meteorological disasters under complex terrain, which is of great significance to the development of related reanalysis data. In addition, the idea of ​​improving the observation operator in the embodiment of the present invention is universal. In the future, the method can be promoted for direct assimilation of more series of satellites and satellite-borne microwave instruments, further promoting the actual application rate of satellite data. In view of the technical difficulty of assimilating satellite-borne microwave observations that are sensitive to the ground surface under current complex terrain, the embodiments of the present invention innovatively utilize the land surface temperature products of the new generation Fengyun-4B satellite and machine learning methods to develop a set of background field land surface temperature correction algorithms with high timeliness, and build corresponding observation operator optimization modules inside the existing mainstream data assimilation system (WRFDA), which can support the real-time assimilation of Fengyun microwave hygrometer data in the numerical forecast process of different underlying surfaces or regions, and significantly improve the application rate and assimilation quality of the FY-3 polar-orbiting satellite microwave hygrometer data. The method is convenient and efficient, and can be used for the direct assimilation of different satellite-borne microwave hygrometers in the future, making it possible to extract a large amount of land information from satellite-borne microwave observations.

[0055] 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 referenced to each other.

[0056] Figure 7 A schematic diagram of the structure of a Fengyun satellite observation data assimilation device provided by an embodiment of the present invention is shown in FIG. Figure 7 As shown, the Fengyun satellite observation data assimilation device provided by the embodiment of the present invention includes: A generating module 701 is used to generate a numerical model background field using a regional model; The alignment module 702 is used to obtain the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, align the real-time land surface temperature with the spatial grid points of the numerical model background field in time and space resolution, and obtain the measured area of ​​the land surface temperature matching the spatial grid points of the background field; 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 area if the measured area is a non-missing area; if the measured area is a missing area, correct the land surface temperature at the corresponding time in the background field based on the land surface temperature correction model to obtain an optimized background field; A construction module 704 is used to construct an error 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; The acquisition module 705 is used to collect the original radiation rate observation data collected by the second satellite observation device, and pre-process the original radiation rate observation data to obtain the observed radiation rate; An 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 an optimal analysis field; The land surface temperature correction model is trained based on the land surface temperature inverted from history and the meteorological element characteristics simulated by the regional model of the corresponding time period.

[0057] The Fengyun satellite observation data assimilation device provided in an embodiment of the present invention generates a numerical model background field by using a regional model; obtains the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, aligns the time and space resolution of the real-time land surface temperature with the spatial grid points of the numerical model background field, and obtains 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, 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 at the corresponding time in the background field are corrected based on the land surface temperature correction model to obtain an optimized background field; based on the historical background field data, an error describing each variable in the background field is constructed to obtain background error information. , converting the optimized background field and the background error information into simulated radiance; collecting the original radiance observation data collected by the second satellite observation equipment, preprocessing the original radiance observation data to obtain the observed radiance; combining the observed radiance with the optimized background field to obtain the analysis field, iteratively optimizing the analysis field, and outputting the optimal analysis field; wherein the land surface temperature correction model is trained based on the land surface temperature obtained by historical inversion and the meteorological element characteristics simulated by the regional model of the corresponding time period. The present 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, so that the system has better satellite data assimilation capabilities, and at the same time improves the assimilation effect of satellite data under high-altitude complex terrain.

[0058] Figure 8 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 8As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The memory 830 includes a computer program, an operating system and acquired data. The processor 810 may call the logic instructions in the memory 830 to execute the Fengyun satellite observation data assimilation method, which includes: generating a numerical model background field using a regional model; acquiring the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, aligning the real-time land surface temperature with the spatial grid of the numerical model background field in time and space resolution, and obtaining 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 model based on the land surface temperature The land surface temperature at the corresponding moment in the background field is corrected to obtain an optimized background field; based on the historical background field data, an error describing each variable in the background field is constructed to obtain background error information, and the optimized background field and the background error information are converted into simulated emissivity; the original emissivity observation data collected by the second satellite observation device are collected, and the original emissivity observation data are preprocessed to obtain an observed emissivity; the observed emissivity is combined with the optimized background field to obtain an analysis field, and the analysis field is iteratively optimized to output an optimal analysis field; wherein the land surface temperature correction model is trained based on the land surface temperature derived from historical inversion and the characteristics of meteorological elements simulated by the regional model of the corresponding time period.

[0059] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, or the part that contributes to the relevant technology or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0060] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the Fengyun satellite observation data assimilation method provided by the above-mentioned methods, the method comprising: generating a numerical model background field using a regional model; obtaining the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, aligning the real-time land surface temperature with the spatial grid points of the numerical model background field in time and space resolution, and obtaining 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 moment with the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, based on the land surface temperature The land surface temperature at the corresponding moment in the background field is corrected by a correction model to obtain an optimized background field; the error describing each variable in the background field is constructed based on historical background field data to obtain background error information, and the optimized background field and the background error information are converted into simulated emissivity; the original emissivity observation data collected by the second satellite observation device are collected, and the original emissivity observation data are preprocessed to obtain observed emissivity; the observed emissivity is combined with the optimized background field to obtain an analysis field, and the analysis field is iteratively optimized to output an optimal analysis field; wherein the land surface temperature correction model is trained based on the land surface temperature obtained by historical inversion and the meteorological element characteristics simulated by the regional model of the corresponding time period.

[0061] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0062] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions 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 numerical model background field using regional model; Acquire the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, align the real-time land surface temperature with the spatial grid points of the numerical model background field in time and space resolution, and obtain the measured area of ​​the land surface temperature matching the spatial grid points of the background field; If the measured area is a non-missing area, the land surface temperature in the background field at the corresponding time is replaced by the real-time land surface temperature corresponding to the non-missing area; if the measured area is a missing area, the land surface temperature at the corresponding time in the background field is corrected based on the land surface temperature correction model to obtain an optimized background field; Based on historical background field data, an error describing each variable in the background field is constructed to obtain background error information, and the optimized background field and the background error information are converted into simulated radiance; Collecting original radiation rate observation data collected by the second satellite observation device, and preprocessing the original radiation rate observation data to obtain observed radiation rate; Combining the observed radiance with the optimized background field to obtain an analysis field, iteratively optimizing the analysis field, and outputting an optimal analysis field; The land surface temperature correction model is trained based on the land surface temperature inverted from history and the meteorological element characteristics simulated by the regional model of the corresponding time period.

2. The Fengyun satellite observation data assimilation method according to claim 1, characterized in that: The correcting the land surface temperature at the corresponding time in the background field based on the land surface temperature correction model includes: Extract the characteristic vectors of meteorological elements closely related to land surface temperature from numerical models; Collect the land surface temperature data inverted in the historical time period and the characteristic data of meteorological elements simulated by the regional model in the corresponding time period to construct a training set; The random forest algorithm is used to fit the training samples in the training set to obtain a land surface temperature correction model; The meteorological element characteristic vector is input into the land surface temperature correction model to obtain the predicted land surface temperature 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 radiation rate observation data to obtain the observed radiation rate includes: filtering the original radiance observation data; The observed radiance data after filtering is corrected using the dynamic deviation correction method to obtain the observed radiance.

4. The Fengyun satellite observation data assimilation method according to claim 1, characterized in that: The error of each variable in the background field is constructed based on the historical background field data to obtain background error information, including: Based on the historical background field data of the study area, key variables in the model background field are extracted; Using the background error calculation module in the WRFDA assimilation system, the error values ​​of the key variables and their spatial correlation are calculated; Based on the error values ​​and spatial correlation, a background error covariance matrix is ​​constructed, wherein the covariance matrix is ​​used to describe the error distribution of each variable in the model background field and their mutual relationship; The background error covariance matrix is ​​standardized by the WRFDA assimilation system to generate background error information.

5. The Fengyun satellite observation data assimilation method according to claim 1, characterized in that: The step of converting the optimized background field and the background error information into simulated radiance comprises: According to the type of the second satellite observation device, matching the coefficient file of the radiation transfer model; The coefficient file, the optimized background field and the background error information are input into an observation operator, and the background field variables are converted into simulated radiance through the observation operator.

6. The method for assimilating Fengyun satellite observation data according to claim 1, characterized in that: After obtaining the observed radiance, it also includes: Calculating the difference between the observed radiance and the simulated radiance to obtain an observed increment; Eliminate observation points on the spatial grid whose observation increment is greater than the preset threshold, and select the observed radiances for subsequent assimilation; The screened observed radiance is combined with the optimized background field to obtain the analysis field.

7. The method for assimilating Fengyun satellite observation data according to claim 6, characterized in that: The iterative optimization of the analysis field to output the optimal analysis field comprises: Combining the screened observed radiance with the optimized background field using a three-dimensional variational method, constructing an objective function, and calculating an analysis field based on the objective function; The optimized background field is replaced by the analysis field, and the field is recombined with the screened observed radiance. The objective function is iteratively calculated to minimize the objective function, thereby obtaining the optimal analysis field.

8. A Fengyun satellite observation data assimilation device, characterized in that: include: A generation module, used for generating a numerical model background field using a regional model; An alignment module is used to obtain the real-time land surface temperature inverted by the thermal infrared channel of the first satellite observation device, align the real-time land surface temperature with the spatial grid points of the numerical model background field in time and space resolution, and obtain a measured area of ​​the land surface temperature matching the spatial grid points of the background field; an optimization module, for 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 non-missing area; and for correcting the land surface temperature at the corresponding time in the background field based on the land surface temperature correction model if the measured area is a missing area, so as to obtain an optimized background field; A construction module, used to construct an error 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; A collection module, used to collect the original radiation rate observation data collected by the second satellite observation device, and pre-process the original radiation rate observation data to obtain the observed radiation rate; An output module, used for combining the observed radiance with the optimized background field to obtain an analysis field, iteratively optimizing the analysis field, and outputting an optimal analysis field; The land surface temperature correction model is trained based on the land surface temperature inverted from history and the meteorological element characteristics simulated by the regional model of the corresponding time period.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the Fengyun satellite observation data assimilation method as described in any one of claims 1 to 7 is implemented.

10. A non-transitory readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for assimilating Fengyun satellite observation data as described in any one of claims 1 to 7 is implemented.

Images