Method and System for Identifying Heavy Rainfall Areas Based on the Fusion Wind Field of Geostationary Satellite Cloud-Guided Winds

Abstract

The present invention discloses a method and system for identifying heavy rain falling areas based on a fused wind field of geostationary satellite cloud-derived winds. The method includes: obtaining the original cloud-derived wind data of a geostationary meteorological satellite and performing preprocessing to generate high-quality geostationary satellite cloud-derived wind data. Obtaining the numerical model forecast wind field in real time, and fusing the satellite cloud-derived wind data and the numerical model forecast wind field to generate three-dimensional fused wind field data. Based on the three-dimensional fused wind field data, calculating and generating diagnostic physical quantities through a first preset method. Judging the heavy rain falling areas based on the diagnostic physical quantities to generate a first heavy rain falling area. Revising the forecast heavy rain falling area based on the first heavy rain falling area to generate a second heavy rain falling area. Thus, the method for identifying heavy rain falling areas based on a fused wind field of geostationary satellite cloud-derived winds of the present invention realizes accurate forecasting of heavy rain falling areas, thereby improving the timeliness and accuracy of weather forecasting.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of meteorological forecasting, and in particular, to a method and system for identifying heavy rain falling areas based on geostationary satellite cloud-derived wind integrated wind field. Background Art

[0002] In existing weather services, the forecasting of heavy rain falling areas mainly relies on numerical models. However, limited by the initial value conditions and the description of real weather physical processes, the forecasting accuracy is limited, especially for sudden heavy rain events. Although weather radars and conventional sounding data are used in the forecasting of severe convective weather, their spatio-temporal resolution is low and it is difficult to meet the continuous monitoring requirements. Although satellite observations can provide cloud map information for the entire region and at all times, further data processing and integration are still required for direct application in heavy rain falling area forecasting. Summary of the Invention

[0003] The embodiments of the present invention provide a method and system for identifying heavy rain falling areas based on geostationary satellite cloud-derived wind integrated wind field, which realizes accurate forecasting of heavy rain falling areas, thereby improving the timeliness and accuracy of weather forecasting.

[0004] To achieve the above object, in a first aspect, the present invention provides a method for identifying heavy rain falling areas based on geostationary satellite cloud-derived wind integrated wind field, including:

[0005] Obtaining the original cloud-derived wind data of a geostationary meteorological satellite and performing preprocessing to generate high-quality cloud-derived wind data of the geostationary satellite;

[0006] Real-time obtaining the numerical model forecast wind field, and integrating the satellite cloud-derived wind data and the numerical model forecast wind field to generate three-dimensional integrated wind field data;

[0007] Based on the three-dimensional integrated wind field data, calculating and generating diagnostic physical quantities through a first preset method;

[0008] Based on the diagnostic physical quantities, identifying the heavy rain falling areas to generate a first heavy rain falling area;

[0009] Based on the first heavy rain falling area, correcting the forecast heavy rain falling area to generate a second heavy rain falling area.

[0010] In an embodiment of the present invention, the obtaining the original cloud-derived wind data of a geostationary meteorological satellite and performing preprocessing to generate high-quality cloud-derived wind data of the geostationary satellite includes:

[0011] Obtaining the water vapor channel cloud-derived wind information and the infrared channel cloud-derived wind information of the geostationary meteorological satellite;

[0012] Based on the water vapor channel cloud-derived wind information and the infrared channel cloud-derived wind information, integrating to generate the original cloud-derived wind data;

[0013] Perform quality control on the original cloud-derived wind data to generate high-quality cloud-derived wind data for geostationary satellites.

[0014] In an embodiment of the present invention, the real-time acquisition of the numerical model forecast wind field, and the fusion of the satellite cloud-derived wind data and the numerical model forecast wind field to generate three-dimensional fusion wind field data includes:

[0015] Real-time acquire the numerical model forecast wind field;

[0016] Fuse the satellite cloud-derived wind data and the numerical model forecast wind field by a second preset method to generate the three-dimensional fusion wind field data.

[0017] In an embodiment of the present invention, the fusion of the satellite cloud-derived wind data and the numerical model forecast wind field by a second preset method to generate the three-dimensional fusion wind field data includes:

[0018] Calculate the background error covariance matrix B of the numerical model forecast wind field;

[0019] Calculate the observation error covariance matrix R of the satellite cloud-derived wind data;

[0020] Based on the background error covariance matrix B and the observation error covariance matrix R, through the optimal interpolation algorithm, fuse the satellite cloud-derived wind data and the numerical model forecast wind field to generate the three-dimensional fusion wind field data.

[0021] In a second aspect, the present invention provides a heavy rainfall area identification system based on a geostationary satellite cloud-derived wind fusion wind field, including: a first generation module, a second generation module, a third generation module, a fourth generation module, and a fifth generation module. The first generation module is used to acquire the original cloud-derived wind data of the geostationary meteorological satellite, and perform preprocessing to generate high-quality cloud-derived wind data for geostationary satellites. The second generation module is used to real-time acquire the numerical model forecast wind field, and fuse the satellite cloud-derived wind data and the numerical model forecast wind field to generate three-dimensional fusion wind field data. The third generation module is used to calculate and generate diagnostic physical quantities based on the three-dimensional fusion wind field data by a first preset method. The fourth generation module is used to perform heavy rainfall area identification based on the diagnostic physical quantities to generate a first heavy rainfall area. And the fifth generation module is used to revise the forecast heavy rainfall area based on the first heavy rainfall area to generate a second heavy rainfall area.

[0022] In an embodiment of the present invention, the first generation module includes: a first acquisition unit, a first generation unit, and a second generation unit. The first acquisition unit is used to acquire water vapor channel cloud-derived wind information and infrared channel cloud-derived wind information of a geostationary meteorological satellite. The first generation unit is used to integrate the water vapor channel cloud-derived wind information and the infrared channel cloud-derived wind information to generate the original cloud-derived wind data. The second generation unit is used to perform quality control on the original cloud-derived wind data to generate high-quality cloud-derived wind data of the geostationary satellite.

[0023] In an embodiment of the present invention, the second generation module includes: a second acquisition unit and a third generation unit. The second acquisition unit is used to acquire the numerical model forecast wind field in real time. And the third generation unit is used to fuse the satellite cloud-derived wind data and the numerical model forecast wind field by a second preset method to generate the three-dimensional fused wind field data.

[0024] In an embodiment of the present invention, the third generation unit includes:

[0025] A first calculation sub-unit, configured to calculate the background error covariance matrix B of the numerical model forecast wind field;

[0026] A second calculation sub-unit, configured to calculate the observation error covariance matrix R of the satellite cloud-derived wind data;

[0027] A generation sub-unit, configured to fuse the satellite cloud-derived wind data and the numerical model forecast wind field based on the background error covariance matrix B and the observation error covariance matrix R through an optimal interpolation algorithm to generate the three-dimensional fused wind field data.

[0028] At least one processor; and

[0029] A memory communicatively connected to the at least one processor;

[0030] Wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for identifying heavy rain falling areas based on geostationary satellite cloud-derived wind fused wind fields as described above.

[0031] Fourthly, the present invention provides a computer-readable storage medium, including a computer program and instructions, when the computer program or the instructions run on a computer, enabling the computer to execute the method for identifying heavy rain falling areas based on geostationary satellite cloud-derived wind fused wind fields as described above.

[0032] Compared with the prior art, the method and system for identifying heavy rain falling areas based on geostationary satellite cloud-derived wind fused wind fields according to the present invention have the following beneficial effects:

[0033] 1. The adaptive intelligent optimal interpolation method can dynamically adjust the interpolation coefficient according to real-time data. Combining with the accurate estimation of observation error by the neural network, it significantly improves the accuracy of wind field fusion.

[0034] 2. It is applicable to the data fusion requirements of different atmospheric vertical levels and provides stable and reliable interpolation results under rapidly changing meteorological conditions.

[0035] 3. The fusion process is rapid, meeting the business timeliness requirements of weather forecasting, and can quickly generate three-dimensional fusion wind field data with high spatio-temporal resolution.

[0036] 4. It not only provides the optimal three-dimensional wind field gridded product, but also provides diagnostic physical quantity products such as vorticity, divergence, and vertical wind shear, providing more comprehensive decision-making support for forecasters.

[0037] 5. By monitoring the severe convective activities in the potential heavy rainfall areas, it effectively corrects the heavy rainfall areas predicted by the model and improves the accuracy and reliability of the forecast. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a schematic flowchart of a method for identifying heavy rainfall areas based on the fusion of geostationary satellite cloud winds in the first embodiment of the present invention;

[0039] Figure 2 is a schematic structural diagram of a system for identifying heavy rainfall areas based on the fusion of geostationary satellite cloud winds in the second embodiment of the present invention;

[0040] Figure 3 is a schematic structural diagram of an electronic device in the third embodiment of the present invention;

[0041] Figure 4 is a schematic logical flowchart of a method for identifying heavy rainfall areas based on the fusion of geostationary satellite cloud winds in a specific embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following further elaborates on the embodiments of the present invention in conjunction with the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention, rather than limiting the embodiments of the present invention. Additionally, it should be noted that for the sake of description, only parts related to the embodiments of the present invention are shown in the drawings rather than all structures.

[0043] For the convenience of understanding, the main implementation concepts of the embodiments of the present invention are first briefly described.

[0044] In current weather services, the prediction of heavy rain areas mainly relies on numerical models. With the increasing improvement of numerical prediction technology and the rapid increase in computing power, the accuracy of numerical model forecasts for weather has been continuously improving. However, the model is restricted by initial value conditions and the description of real weather physical processes, making the prediction of heavy rain still the focus and difficulty of current weather forecasting. Especially for sudden heavy rain processes, there are significant deviations in the prediction of heavy rain areas by multi - models at home and abroad, seriously threatening people's lives and property safety. Therefore, we cannot solely rely on numerical models to predict heavy rain areas.

[0045] Although weather radars have been widely used in the nowcasting of severe convective weather, when the weather radar captures precipitation echoes, the precipitation has already occurred. Conventional sounding data is obtained by sounding balloons to detect various meteorological elements from top to bottom in the atmosphere. However, the spatio - temporal resolution of conventional sounding data is relatively low (twice a day, and it is impossible to set up stations in special terrain areas), and it cannot meet the continuous monitoring of weather events.

[0046] The inventor, by discovering the technical defects described in the above - mentioned background art, fuses the cloud - derived wind of geostationary meteorology and the information of the numerical model forecast wind field to form three - dimensional fused wind field data with high spatio - temporal resolution. The fused wind field can well reflect the characteristics of synoptic - scale and mesoscale weather systems, and at the same time conforms to the usage habits of forecasters. It can also effectively correct the numerical model wind field. Meanwhile, the fusion method is rapid and meets the requirements of the timeliness of weather forecasting operations. After fusion, it can not only provide the optimal fine three - dimensional wind field gridded products, but also provide relevant physical quantity diagnostic products, ultimately improving the accuracy of heavy rain area prediction.

[0047] Embodiment 1 Figure 1 is a schematic flowchart of a method for identifying heavy rain areas based on the fusion of geostationary satellite cloud - derived wind fields in Embodiment 1 of the present invention. As Figure 1 shown, Embodiment 1 provides a method for identifying heavy rain areas based on the fusion of geostationary satellite cloud - derived wind fields, including:

[0048] Step S100: Obtain the original cloud - derived wind data of geostationary meteorological satellites and perform pre - processing to generate high - quality cloud - derived wind data of geostationary satellites;

[0049] Specifically, collect the original cloud - derived wind data of water vapor channels and infrared channels from geostationary meteorological satellites. By integrating this information, a comprehensive cloud - derived wind data set is constructed. Subsequently, a strict quality control process is adopted to screen the original data, eliminate low - quality data, and retain high - quality information. This series of pre - processing operations ensures the accuracy and reliability of the data, and finally generates high - quality cloud - derived wind data of geostationary satellites for subsequent analysis, laying a solid foundation for subsequent wind field fusion and heavy rain area identification.

[0050] Step S200, obtaining the wind field predicted by the numerical model in real time, and fusing the satellite cloud wind guide data and the wind field predicted by the numerical model to generate three-dimensional fused wind field data;

[0051] Specifically, for example, the numerical model forecast wind field data is accurately obtained from the meteorological bureau's data acquisition platform, and these data provide a preliminary prediction of the atmospheric wind field. Next, the advanced adaptive intelligent optimal interpolation method is used to deeply integrate the satellite cloud wind data with the numerical model forecast wind field. This process fully considers the error characteristics of the observation field and the background field, and dynamically adjusts the interpolation coefficient through an intelligent algorithm to ensure that the error of the fused wind field data is minimized. Finally, three-dimensional fused wind field data with high temporal and spatial resolution is generated. This data not only retains the macro-prediction ability of the numerical model, but also incorporates the fine observation information of the satellite cloud wind, providing strong support for the accurate identification of the rainstorm area.

[0052] Step S300, based on the three-dimensional fused wind field data, a diagnostic physical quantity is calculated and generated by a first preset method;

[0053] Specifically, based on the three-dimensional fused wind field data generated in step S200, a pre-set first preset method is used to perform in-depth calculations and analyses. In this process, by carefully processing key parameters such as wind speed and wind direction in the fused wind field, important diagnostic physical quantities such as vorticity, divergence, and vertical wind shear are calculated. These physical quantities not only reveal the inherent laws of atmospheric motion, but also provide a key basis for the subsequent identification of rainstorm areas. Through the calculation and generation of step S300, more comprehensive and accurate wind field diagnostic information is provided to weather forecasters.

[0054] Step S400, identifying a rainstorm area based on the diagnostic physical quantity to generate a first rainstorm area;

[0055] Specifically, by comprehensively analyzing the spatial distribution and variation trends of these physical quantities, and combining meteorological principles and empirical models, the areas where heavy rain is likely to occur are accurately identified. This process fully considers the dynamic evolution characteristics of weather systems and the indicative role of different physical quantities in the formation of heavy rain, and finally generates the first heavy rain fall area. This forecast result not only provides an important reference for disaster prevention and mitigation, but also lays a foundation for subsequent model forecast correction. Among them, vorticity is a physical quantity that describes the strength of fluid rotation and is commonly used in meteorology to characterize the intensity of weather systems. Heavy rain is often associated with strong vorticity centers. Analyze and fuse the vorticity distribution in the wind field, especially to find the positive vorticity center (corresponding to the trough area) and the negative vorticity center (corresponding to the ridge area). Strong heavy rain events are usually related to the configuration of negative vorticity in the upper layer and positive vorticity in the lower layer, and this configuration often indicates significant vertical upward movement. Divergence is a physical quantity that measures the expansion or contraction of fluid volume and is particularly important for meteorological forecasts; the positive and negative centers and their distribution patterns in the divergence field are closely related to the distribution of severe convective weather. Analyze and fuse the divergence distribution in the wind field, especially to find the negative centers (convergence areas) in the divergence field. These areas are often high-probability areas where heavy rain occurs. In addition, mesoscale convergence areas often appear prior to precipitation, so monitoring the changes in the divergence field is of great significance for forecasting short-term heavy precipitation. Vertical wind shear refers to the rate of change of wind speed or wind direction with height and is an important factor promoting convective tilt and the formation of severe convective systems (such as supercell storms). Analyze and fuse the vertical wind shear distribution in the wind field, especially to find areas with strong wind shear. These areas are often associated with heavy rainfall or even heavy rain events. In addition, monitoring the variation trend of vertical wind shear within 24 hours can more accurately forecast the time and location of heavy rain. Combine the analysis results of diagnostic physical quantities such as vorticity, divergence, and vertical wind shear, and comprehensively judge the areas where heavy rain is likely to occur. By comparing the spatial distribution and variation trends of different physical quantities, determine the high-probability areas of the heavy rain fall area.

[0056] Step S500, correct the forecast heavy rain fall area based on the first heavy rain fall area to generate a second heavy rain fall area;

[0057] Specifically, step S500 is the refined adjustment stage of heavy rain area prediction. Based on the first heavy rain area result obtained in step S400, it further combines the original heavy rain area information predicted by the numerical model for detailed comparative analysis. By evaluating the differences between the two, taking advantage of the satellite cloud-derived wind integrated wind field, a scientific and reasonable correction is made to the heavy rain area predicted by the model. This process not only considers the complexity and uncertainty of atmospheric motion but also fully utilizes the accuracy of real-time observation data, ultimately generating a more accurate second heavy rain area. This correction result significantly improves the accuracy and reliability of heavy rain area prediction, providing more powerful support for disaster prevention and mitigation work. Among them, the first heavy rain area (identified based on diagnostic physical quantities) is compared and analyzed with the original heavy rain area predicted by the model, comparing the characteristics such as their spatial distribution, range size, and intensity. The differences between the first heavy rain area and the heavy rain area predicted by the model are evaluated, and these differences may stem from the uncertainty of the model initial value conditions, the limitations of physical process description, and the errors of observation data, etc. According to the difference evaluation results, a correction strategy is formulated. If the difference between the first heavy rain area and the model prediction is large and the analysis based on diagnostic physical quantities is more reliable, it is more inclined to use the first heavy rain area as the correction basis. During the correction process, the information of multiple diagnostic physical quantities and their interactions and influences can be comprehensively considered to improve the accuracy and reliability of the correction. According to the correction strategy, the heavy rain area predicted by the model is adjusted and optimized to generate the second heavy rain area. This process involves, for example, adjusting the range, shape, and intensity of the heavy rain area to more accurately reflect the actual weather conditions. The corrected second heavy rain area is verified by comparing its coincidence with the actual observation data. Through continuous verification and feedback, the correction strategy and method are optimized to improve the accuracy and reliability of heavy rain area prediction.

[0058] In an embodiment of the present invention, the step S100 includes:

[0059] Step S101, obtaining the cloud-derived wind information of the water vapor channel and the cloud-derived wind information of the infrared channel of the geostationary meteorological satellite;

[0060] Specifically, step S101 is the initial step of data collection, focusing on capturing key wind field information from the geostationary meteorological satellite. Specifically, this step aims to obtain the cloud-derived wind data of the water vapor channel and the infrared channel of the satellite. The water vapor channel is sensitive to the distribution of water vapor in the atmosphere and helps to identify the humidity changes in the cloud system; while the infrared channel reveals the temperature structure of the cloud top by measuring the infrared radiation emitted by the earth's surface and clouds, and then infers the wind speed and direction. For example, FY-4 (Fengyun-4) is China's new generation of geostationary meteorological satellite, which can provide high spatio-temporal resolution water vapor and infrared cloud images simultaneously, providing a rich and accurate data source for step S101.

[0061] Step S102: Integrate the water vapor channel cloud motion wind information and the infrared channel cloud motion wind information to generate the original cloud motion wind data;

[0062] Specifically, comprehensively integrate the water vapor channel cloud motion wind information obtained in step S101 and the infrared channel cloud motion wind information. For example, if there are 100 observation points in the water vapor channel and also 100 observation points in the infrared channel, during integration, these two sets of data are combined to form a data set containing 200 observation points. This process takes into account the complementarity of the data from the two channels and combines their respective advantages in revealing the dynamic characteristics of cloud systems to generate a more comprehensive and multi-dimensional original cloud motion wind data set. Taking the FY-4 (Fengyun-4) satellite as an example, the multi-channel cloud map data it provides is widely used in cloud motion wind retrieval. By integrating the information of its water vapor and infrared channels, the wind speed and wind direction changes inside the cloud system can be captured more accurately, laying a solid foundation for subsequent wind field analysis and heavy rain prediction.

[0063] Step S103: Perform quality control on the original cloud motion wind data to generate the high-quality geostationary satellite cloud motion wind data;

[0064] Specifically, step S103 is an important link to ensure data accuracy. It implements strict quality control on the original cloud motion wind data generated in step S102. Through a series of automated means, abnormal or low-quality data points caused by instrument errors, cloud occlusion, or atmospheric interference are removed, and high-quality and reliable cloud motion wind information is retained. This process effectively improves the accuracy and credibility of the data, providing a solid foundation for subsequent meteorological analysis and forecasting. Taking the FY-4 (Fengyun-4) satellite as an example, the quality indicator (QI) of its cloud motion wind data is used to represent the data quality of the retrieved cloud motion wind. The larger the value, the higher the accuracy of the cloud motion wind data generally. We consider that the cloud motion wind data with a QI value greater than 80 is high-quality. After the quality control in step S103, it can more accurately reflect the actual situation of the atmospheric wind field and provide strong support for meteorological services such as heavy rain fall area forecasting.

[0065] In an embodiment of the present invention, the step S200 includes:

[0066] Step S201: Obtain the numerical model forecast wind field in real time;

[0067] Specifically, step S201 involves obtaining the latest numerical model forecast wind field data from the meteorological bureau data acquisition platform. The numerical model forecast wind field is the prediction result of the atmospheric wind field in the future for a period of time, which is calculated by high-performance computer simulation based on complex mathematical models and a large amount of observational data. For example, the meteorological bureau regularly releases the global wind field forecast data generated by its integrated forecasting system, and these data are widely used in various meteorological services and scientific research activities. In step S201, the system automatically and real-time downloads these numerical model forecast wind field data from the meteorological bureau information center for subsequent fusion processing with satellite cloud drift wind data.

[0068] Step S202, fuse the satellite cloud drift wind data and the numerical model forecast wind field through a second preset method to generate the three-dimensional fusion wind field data;

[0069] Specifically, the obtained numerical model forecast wind field and satellite cloud drift wind data are deeply integrated by using a preset second preset method. This method usually involves optimal interpolation processing technology, aiming to comprehensively consider the macroscopic prediction ability of the numerical model and the fine structure information of satellite observations, and dynamically adjust the fusion weights through intelligent algorithms to generate three-dimensional fusion wind field data with high spatio-temporal resolution and small errors.

[0070] In an embodiment of the present invention, step S202 includes:

[0071] Step S2021, calculate the background error covariance matrix B of the numerical model forecast wind field;

[0072] Specifically, the background error covariance matrix B is used to describe the statistical correlation between prediction errors at different positions in the numerical model forecast wind field. This matrix contains the variance of the prediction error at each grid point (the elements on the diagonal of the matrix) and the covariance of the prediction errors between different grid points (the off-diagonal elements of the matrix). Calculating the B matrix usually involves statistical analysis of the historical data of the model forecast to determine the prediction error distribution and its spatial correlation at each grid point. This process is crucial for subsequent fusion of satellite cloud drift wind data and numerical model forecast wind field, because it can help determine how to weight data from different sources according to the prediction uncertainty of the numerical model during the fusion process, so as to generate more accurate three-dimensional fusion wind field data.

[0073] Step S2022, calculate the observation error covariance matrix R of the satellite cloud drift wind data;

[0074] Specifically, the observation error covariance matrix R is used to describe the statistical correlation of observation errors between different observation points in the satellite cloud wind data. The R matrix contains the variances of the observation errors at each observation point (the elements on the diagonal of the matrix) and the covariances of the observation errors between different observation points (the elements off the diagonal of the matrix). To calculate the R matrix, it is usually necessary to perform spatio-temporal matching using historical satellite cloud wind data and corresponding radiosonde observation data, extract the observation errors at each observation point (i.e., the differences between the satellite cloud wind values and the radiosonde observation values), and calculate the variances and covariances of these errors through statistical methods. In addition, neural network technology can be used to more accurately estimate the complex structure and spatial correlation of the observation errors. By training a neural network model, it can learn the patterns of different observation errors, thus obtaining more accurate observation error predictions. The finally obtained R matrix not only contains diagonal elements but also non-diagonal elements reflecting the spatial correlation of the observation errors, which is crucial for subsequently fusing satellite cloud wind data and numerical model forecast wind fields through the optimal interpolation algorithm.

[0075] Step S2023, based on the background error covariance matrix B and the observation error covariance matrix R, fuse the satellite cloud wind data and the numerical model forecast wind field through the optimal interpolation algorithm to generate the three-dimensional fused wind field data;

[0076] Specifically, according to the background error covariance matrix B and the observation error covariance matrix R, calculate the weights for each observation point and each numerical model grid point. The weights reflect the relative importance of the observation data and the model data in the fusion process. The magnitudes of the weights depend on the statistical characteristics of the observation errors and the model errors, as well as the spatial correlation between the observation points and the grid points. Using the calculated weights, perform weighted averaging on the satellite cloud wind observation data and the numerical model forecast wind field to generate the fused wind field data. Among them, for each grid point, its fused wind field value will be the result of weighted averaging of the observation values and the model forecast values of the surrounding observation points according to the weights. Through the optimal interpolation algorithm, ensure that the fused wind field data has the minimum error in the statistical sense. This is achieved by minimizing a cost function that takes into account the statistical characteristics of the observation errors and the model errors, as well as the spatial correlation between the observation data and the model data. Interpolate the fused wind field data onto a regular three-dimensional grid to generate the three-dimensional fused wind field data. This step ensures that the fused wind field data has a unified spatial resolution and format, facilitating subsequent analysis and application.

[0077] Through the above steps, step S2023 can make full use of the advantages of the satellite cloud wind data and the numerical model forecast wind field to generate high-precision, high spatio-temporal resolution three-dimensional fused wind field data, providing strong support for the accurate forecast of the heavy rain fall area.

[0078] In an embodiment of the present invention, the step S300 includes:

[0079] Based on the three-dimensional fused wind field data, calculate vorticity, divergence, vertical wind shear, and 24-hour shear trend to generate the diagnostic physical quantities;

[0080] Among them, the formula for vorticity calculation is:

[0081]

[0082] Among them, the formula for divergence calculation is:

[0083]

[0084] Among them, the formula for vertical wind shear is:

[0085]

[0086] Among them, u is the zonal wind speed, v is the meridional wind speed, a is the radius of the earth, φ is the latitude, j is the grid point index in the latitude direction, and i is the grid point index in the longitude direction.

[0087] Among them, rv(j,i) is the vorticity value, which is a rotation measure calculated from the u and v wind components; dv(j,i) is the divergence value, which is a divergence measure calculated from the u and v wind components; dy2(j) is the grid spacing in the latitude direction, dx2(j) is the grid spacing in the longitude direction; s is the vertical wind shear value, which is the wind shear intensity calculated from the wind speed difference between different height layers; tan is the tangent function, and tan(φ(j)) is the tangent value of the latitude of the jth grid point;

[0088] Among them, the 24-hour shear trend is the difference in the vertical wind shear over 24 hours.

[0089] Figure 4 It is a schematic diagram of the logical process of the heavy rain fall area identification method based on the geostationary satellite cloud-derived wind fused wind field in a specific embodiment of the present invention. As Figure 4 shown, in a specific embodiment, the heavy rain fall area identification method based on the geostationary satellite cloud-derived wind fused wind field of the present invention includes:

[0090] Step 1: Preprocessing of multi-channel information of the geostationary meteorological satellite cloud-derived wind;

[0091] Specifically, the geostationary meteorological satellite contains cloud drift wind information in water vapor channels and infrared channels. To obtain more wind field information, the cloud drift wind data of different channels are integrated. For example, there are 100 observation points in the water vapor channel and 100 observation points in the infrared channel. When integrating, these two sets of data are combined to form a data set containing 200 observation points, from which the cloud drift wind information of all channels can be obtained. Then, the data is marked with an identification code representing the quality of the satellite product. To ensure the fusion accuracy, high-quality satellite cloud drift wind data is selected; the cloud drift wind data is quality-controlled to obtain the quality identification code QI of the cloud drift wind file. Those with a QI greater than 80 are considered cloud drift winds with relatively high retrieval quality and are retained. Considering that both the numerical model and conventional sounding data contain standard pressure level data commonly used by forecasters (50 hPa, 100 hPa, 150 hPa, 200 hPa, 300 hPa, 500 hPa, 700 hPa, 850 hPa), and since the satellite's detection method is from top to bottom and the inversion calculation is carried out based on the moving vector of the cloud, the obtained satellite cloud drift wind data is similar to a scatter distribution, with more cloud drift wind data at high altitudes and less at low altitudes. Therefore, the distribution position of the available cloud drift wind data is not the standard pressure level commonly used by forecasters. To fuse with the model wind field data later, the cloud drift wind data within ±50 hPa of the standard pressure level is searched for, and the quality-controlled satellite cloud drift wind data is processed layer by layer for stratification.

[0092] Step 2: Fuse the cloud drift wind of the geostationary meteorological satellite and the numerical model forecast wind field to construct three-dimensional refined grid wind field data;

[0093] Specifically, the present invention uses an adaptive intelligent optimal interpolation method to fuse the satellite cloud drift wind and the numerical model forecast wind field. The optimal interpolation method can interpolate the observed values at any position to a regular grid and uses the background field as the first guess value. There are errors in both the observed field and the background field, and this method takes into account the expected differences to fuse the observed field and the background field. The error of the fused wind field is the smallest, less than the error of any data in the cloud drift wind or the numerical model product. The optimal interpolation method is a data fusion method that is widely used and relatively mature internationally. It can consider the statistical characteristics of the background field and the observation error, that is, it contains the internal relationship between observation, forecast, and analysis, and determines the best weight in the least squares sense, so that the analysis error can reach the minimum in a statistical sense. The basic principle of this method is that the analysis value at the grid point is obtained by adding the revised value to the background value of the grid point. The revised value is obtained by weighting the differences between the observed values and the background values of the surrounding stations. The weight coefficient (i.e., the optimal interpolation coefficient) should minimize the error of the grid point analysis value. The general form of optimal interpolation is:

[0094] x a =xb +K(y - Hx b (Equation 1-1)

[0095] K = BH T (HBH T +R) -1 (Equation 1-2)

[0096] where x a represents the fused analysis field; x b represents the background field, generally taken as the model forecast value at the previous time step; y represents the observation field; H is the linear observation operator, representing the transformation from the model space to the observation space; B represents the background error covariance matrix; R represents the observation error covariance matrix, and the observation error includes the representativeness error of the observation data. The superscript "T" and "-1" represent the transpose and inverse of the matrix, respectively.

[0097] Optimal interpolation is more formally similar to three-dimensional variational. It is difficult to directly calculate the analysis value from the above equation because it involves the inversion of the B matrix. For high-resolution numerical models, the B matrix is a huge-dimensional matrix, and direct inversion is not practical. However, it can be obtained by minimizing the following cost function J:

[0098]

[0099] Let x = x - x b , y = y - Hx b , substitute into (Equation 1-3), and using the linear observation operator assumption, we can obtain:

[0100]

[0101] where X represents the difference between the analysis field and the background field, also known as the analysis increment; Y represents the difference between the observation field and the background field. That is, by mathematical means (such as the iterative method) to find an appropriate analysis field to minimize the functional J. This method facilitates the use of all observation data in the space for analysis, and compared with other more complex methods (such as four-dimensional variational and ensemble Kalman filtering), it consumes less computer resources.

[0102] It can be seen from (Equation 1-3) that B and R play important roles in the analysis process. At present, there is no exact formula to directly calculate B and R, and only approximate methods can be used for estimation.

[0103] For the estimation of the background error covariance B matrix, use the following formula:

[0104]

[0105] where, It represents the sum of squares of errors (the difference between the grid point value and the true value at that point) of the j-th grid point and is the diagonal element of the B matrix. It represents the statistical variance of the j-th grid point. The error term in the statistical variance is also called the anomaly term, which is the difference between each sample value (grid point value) and the average value within the statistical time period, so that the estimation of the diagonal elements of the background error covariance matrix only depends on its statistical variance and parameter c j .

[0106] In the traditional optimal interpolation algorithm, c j is a fixed value, which may lead to insufficient interpolation accuracy at different altitude levels. By introducing an adaptive algorithm, the value range of c is dynamically adjusted j so that it can be adjusted according to different pressure altitude levels. Through the statistical method of regression analysis, the sounding data and the model background field data at different pressure altitude levels are calculated, and the error distribution and statistical variance of each layer are calculated, so as to estimate the optimal c value of each altitude layer respectively j such that Equation 1-5 becomes the following 1-6

[0107]

[0108] where c j (h) represents the parameter c at different altitude levels h j .

[0109] For the estimation of the off-diagonal elements of the B matrix, the traditional Gaussian distribution form is adopted:

[0110]

[0111] where i and j represent the serial numbers of the grid points, and Δx and Δy respectively represent the horizontal east-west distance and north-south distance between the i-th grid point and the j-th grid point. It is assumed here that the horizontal correlation of the forecast error (background error) decreases exponentially with the increase of the horizontal distance. Lx and Ly are the horizontal correlation scales of the forecast error in the east-west and north-south directions respectively, representing the horizontal distance when the correlation coefficient of the forecast error is reduced to 1 / e, and usually taking constants. In order to improve the accuracy of the estimation, according to different pressure altitude levels and background field data, the regression analysis method is used to dynamically adjust the correlation scales Lx and Ly.

[0112] When calculating the background error covariance matrix B in the present invention, instead of using a fixed constant, an adaptive method is adopted to dynamically adjust the constants at different pressure altitude levels, realizing the estimation of the adaptive dynamic B matrix. This method calculates the errors between the model background field data and the sounding observation data at different pressure altitude levels through regression analysis and statistical methods, so as to optimize the estimation of the B matrix.

[0113] The observation error covariance matrix R represents the observation error of satellite cloud wind data. Compared with the background error covariance matrix B of the numerical model, the R matrix is calculated through neural network technology to more accurately capture the complex structure and spatial correlation of the observation error.

[0114] Using historical satellite cloud wind data and corresponding radiosonde observation data, performing spatio-temporal matching on them, then extracting features (geographical location, time, different pressure altitude levels) and calculated observation errors (i.e., the difference between the satellite cloud wind value and the radiosonde value) from these data to construct a training dataset. Select a neural network model (such as a convolutional neural network), and use the extracted features and observation errors to train the model so that it can learn the patterns of different observation errors. Through the trained neural network model, perform error prediction on real-time satellite cloud wind data to obtain the error variance of each observation point. The diagonal elements of matrix R are the error variances of each observation point (obtained through the convolutional neural network model). Specifically, obtain the cloud wind data of geostationary meteorological satellites from historical archives, including data from water vapor channels and infrared channels. These data should include geographical location (longitude, latitude), time, and wind speed and direction information at different pressure altitude levels. Obtain the corresponding radiosonde observation data for the same period from meteorological observation stations. These data usually include temperature, humidity, wind speed, and wind direction at different pressure altitude levels. Match the satellite cloud wind data with the radiosonde observation data according to time and geographical location to ensure that each pair of matching data points has similar time and spatial positions. If there are time or space mismatches, the data can be adjusted by interpolation or extrapolation methods to improve the matching accuracy. For each pair of matching data points, calculate the difference between the satellite cloud wind value and the radiosonde observation value, and this difference is the observation error. The observation error can be expressed as wind speed error and wind direction error, or synthesized into a comprehensive error value. Extract features from the matching dataset, and these features will be used to train the neural network model. Features can include, for example:

[0115] Geographical location: longitude, latitude;

[0116] Time: year, month, day, hour, minute;

[0117] Pressure altitude level: data at different pressure levels (such as 500 hPa, 850 hPa, etc.);

[0118] Observation error: wind speed error, wind direction error or comprehensive error.

[0119] Combine the extracted features and the corresponding observation errors into training samples. Each training sample should contain input features (geographical location, time, pressure altitude level) and output labels (observation errors), and organize all training samples into a dataset for subsequent training of the neural network model.

[0120] Select an appropriate neural network model according to the complexity of the problem and the characteristics of the data. Considering that the wind field elements have obvious spatial correlation, a convolutional neural network model is selected. Divide the training data set into a training set and a validation set (or test set). Use the training set to train the neural network model, and adjust the model parameters through the backpropagation algorithm to minimize the prediction error. During the training process, use the validation set to monitor the performance of the model to prevent overfitting. Methods such as cross-validation can be used to further optimize the model parameters. The trained neural network model can be used to predict the error of real-time satellite cloud wind data; for each real-time observation point, extract its geographical location, time, pressure altitude layer and other features, and input these features into the neural network model to obtain the predicted observation error (i.e., error variance). Use the predicted error variance to fill the diagonal elements of matrix R; for the off-diagonal elements (representing the error covariance between different observation points), the neural network model can be used to directly predict these covariance values; and determine the optimal spatial correlation model parameters or neural network structure through methods such as cross-validation. Through the above steps, a method for constructing a training data set through historical satellite cloud wind data and radiosonde observation data and using a neural network model for error prediction can be specifically implemented, so as to calculate the observation error covariance matrix R, providing an important basis for subsequent wind field fusion.

[0121] The off-diagonal element represents the error covariance between the i-th point and the j-th point, and the specific formula is: Rij = σ i σ j f, where f is the correlation coefficient predicted by the convolutional neural network model, used to describe the spatial correlation between observation errors. The number of affected observation points is automatically selected as the optimal number of observation points through the method of cross-validation, that is, during the training process, the model is trained and validated according to different numbers of observation points, and the number of observation points with the smallest validation error is selected. In this way, the constructed R matrix not only contains diagonal elements but also off-diagonal elements reflecting spatial correlation, thus more accurately describing the statistical characteristics and spatial correlation of observation errors. This method can capture the complex structure of observation errors and improve the accuracy of data fusion.

[0122] Step 3: Calculate diagnostic physical quantities such as vorticity, divergence, and vertical wind shear according to the fused wind field;

[0123] Specifically, according to the fused wind field information, calculate diagnostic physical quantities such as vorticity, divergence, vertical wind shear, and 24-hour shear trend. u and v represent the zonal wind speed and meridional wind speed respectively, a is the radius of the earth, and φ is the latitude.

[0124] Vorticity formula:

[0125]

[0126] Divergence formula:

[0127]

[0128] Vertical wind shear:

[0129] Among them, rv(j, i): refers to the vorticity value, which is a measure of rotation calculated from the u and v wind components. dv(j, i): refers to the divergence value, which is a measure of divergence calculated from the u and v wind components. s: refers to the vertical wind shear value, which is the intensity of wind shear calculated from the wind speed difference between different altitude levels (200 hPa and 850 hPa). j: refers to the grid point index in the latitude direction. i: refers to the grid point index in the longitude direction. dy2(j): refers to the grid spacing in the latitude direction. dx2(j): refers to the grid spacing in the longitude direction. tan: refers to the tangent function. tan(φ(j)): refers to the tangent of the latitude of the j-th grid point; among them, the 24-hour shear trend is the difference in vertical wind shear over 24 hours.

[0130] Step 4: Based on the diagnostic physical quantities, identify the heavy rain fall area and correct the heavy rain fall area predicted by the model.

[0131] Specifically, the satellite cloud-derived wind contains a very rich amount of data in the upper atmosphere, which is related to the detection principle of the satellite. Therefore, the upper-level wind field of the satellite cloud-derived wind can effectively correct the upper-level wind field predicted by the model. The merged upper-level wind field and its related diagnostic physical quantities, vorticity refers to the vertical relative vorticity, which is commonly used to characterize the intensity of the system. According to synoptic principles, a negative vorticity center corresponds to a ridge area, and a positive vorticity center corresponds to a trough area; the configuration of negative vorticity in the upper level and positive vorticity in the lower level often reflects a relatively significant vertical upward movement. The positive and negative centers and their distribution patterns of the divergence field are closely related to the distribution of severe convective weather. The movement direction of the precipitation area is very consistent with the convergence area, and the mesoscale convergence area often appears 1 - 2 hours before precipitation. Therefore, mastering the changes in the mesoscale divergence field is an important basis for predicting future short-term mesoscale precipitation and the occurrence of heavy rain. Vertical wind shear can promote the tilting of convection and contribute to the formation of strong convective systems such as supercell storms, which are often accompanied by heavy rainfall or even heavy rain. An enhanced wind shear trend is often related to strong convective activities, which may bring extreme rainfall events. By monitoring the 24-hour shear trend, the time and location of heavy rain can be predicted more accurately, improving the accuracy of the prediction. It can well reflect the position of the strong center of convective weather distribution, which is closely related to the position of the heavy rain fall area. The forecaster can compare with the model wind field and diagnostic quantities to correct the heavy rain fall area predicted by the model, effectively improving the prediction accuracy of the heavy rain fall area. Among them, diagnostic variables have certain indicative significance for the diagnosis of heavy rain fall areas in meteorology. For example, the divergence field represents a strong radiation center, indicating that there is strong convective activity below, which may be the position of the heavy rain fall area.

[0132] Embodiment 2 Figure 2 It is a schematic structural diagram of a heavy rain falling area identification system based on geostationary satellite cloud-derived wind integrated wind field in Embodiment 2 of the present invention. As Figure 2 shown, Embodiment 2 provides a heavy rain falling area identification system based on geostationary satellite cloud-derived wind integrated wind field, including: a first generation module, a second generation module, a third generation module, a fourth generation module, and a fifth generation module. The first generation module is used to obtain the original cloud-derived wind data of the geostationary meteorological satellite, perform preprocessing, and generate high-quality geostationary satellite cloud-derived wind data. The second generation module is used to obtain the numerical model forecast wind field in real time, and fuse the satellite cloud-derived wind data and the numerical model forecast wind field to generate three-dimensional integrated wind field data. The third generation module is used to calculate and generate diagnostic physical quantities based on the three-dimensional integrated wind field data through a first preset method. The fourth generation module is used to identify the heavy rain falling area based on the diagnostic physical quantities to generate a first heavy rain falling area. And the fifth generation module is used to correct the forecast heavy rain falling area based on the first heavy rain falling area to generate a second heavy rain falling area.

[0133] In an embodiment of the present invention, the first generation module includes: a first acquisition unit, a first generation unit, and a second generation unit. The first acquisition unit is used to obtain the water vapor channel cloud-derived wind information and the infrared channel cloud-derived wind information of the geostationary meteorological satellite. The first generation unit is used to integrate the water vapor channel cloud-derived wind information and the infrared channel cloud-derived wind information to generate the original cloud-derived wind data. The second generation unit is used to perform quality control on the original cloud-derived wind data to generate the high-quality geostationary satellite cloud-derived wind data.

[0134] In an embodiment of the present invention, the second generation module includes: a second acquisition unit and a third generation unit. The second acquisition unit is used to obtain the numerical model forecast wind field from the meteorological bureau data acquisition platform. And the third generation unit is used to fuse the satellite cloud-derived wind data and the numerical model forecast wind field through a second preset method to generate the three-dimensional integrated wind field data.

[0135] In an embodiment of the present invention, the third generation unit includes:

[0136] a first calculation sub-unit, used to calculate the background error covariance matrix B of the numerical model forecast wind field;

[0137] a second calculation sub-unit, used to calculate the observation error covariance matrix R of the satellite cloud-derived wind data;

[0138] A generation subunit, configured to fuse the satellite cloud wind data and the numerical model forecast wind field based on the background error covariance matrix B and the observation error covariance matrix R through an optimal interpolation algorithm to generate the three-dimensional fused wind field data.

[0139] In an embodiment of the present invention, the third generation module includes a fourth generation unit, configured to calculate vorticity, divergence, vertical wind shear, and 24-hour shear trend based on the three-dimensional fused wind field data to generate the diagnostic physical quantities;

[0140] Among them, the formula for vorticity calculation is:

[0141]

[0142] Among them, the formula for divergence calculation is:

[0143]

[0144] Among them, the formula for vertical wind shear is:

[0145]

[0146] Among them, u is the zonal wind speed, v is the meridional wind speed, a is the radius of the earth, φ is the latitude, j is the grid point index in the latitude direction, and i is the grid point index in the longitude direction.

[0147] Among them, rv(j, i) is the vorticity value, which is a rotation measure calculated from the u and v wind components; dv(j, i) is the divergence value, which is a divergence measure calculated from the u and v wind components; dy2(j) is the grid spacing in the latitude direction, dx2(j) is the grid spacing in the longitude direction; s is the vertical wind shear value, which is the wind shear intensity calculated from the wind speed difference between different altitude layers; tan is the tangent function, and tan(φ(j)) is the tangent of the latitude of the jth grid point;

[0148] Among them, the 24-hour shear trend is the difference in the vertical wind shear over 24 hours.

[0149] The various variations and specific examples of the heavy rain area identification method based on the geostationary satellite cloud wind fused wind field provided in the first embodiment are equally applicable to the heavy rain area identification system based on the geostationary satellite cloud wind fused wind field provided in this embodiment. Through the foregoing detailed description of a heavy rain area identification method based on the geostationary satellite cloud wind fused wind field, those skilled in the art can clearly know the implementation manner of a heavy rain area identification system based on the geostationary satellite cloud wind fused wind field in this embodiment. Therefore, for the sake of brevity of the specification, it will not be elaborated here.

[0150] Example Three Figure 3It is a schematic structural diagram of an electronic device in Embodiment 3 of the present invention. As Figure 3 shown, Embodiment 3 also provides an electronic device 300, which may include: a processor 301 and a memory 302.

[0151] The memory 302 is used to store programs; the memory 302 may include a volatile memory (English: volatile memory), such as a random access memory (English: random-access memory, abbreviation: RAM), such as a static random access memory (English: static random-access memory, abbreviation: SRAM), a double data rate synchronous dynamic random access memory (English: Double Data Rate Synchronous Dynamic Random Access Memory, abbreviation: DDR SDRAM), etc.; the memory may also include a non-volatile memory (English: non-volatile memory), such as a flash memory (English: flash memory). The memory 302 is used to store computer programs (such as application programs and functional modules for implementing the above method), computer instructions, etc. The above computer programs, computer instructions, etc. may be stored in one or more memories 302 in a partitioned manner. And the above computer programs, computer instructions, data, etc. may be called by the processor 301.

[0152] The above computer programs, computer instructions, etc. may be stored in one or more memories 302 in a partitioned manner. And the above computer programs, computer data, etc. may be called by the processor 301.

[0153] The processor 301 is used to execute the computer programs stored in the memory 302 to implement each step in the method involved in the above embodiment.

[0154] Specifically, reference may be made to the relevant descriptions in the foregoing method embodiments.

[0155] The processor 301 and the memory 302 may be in an independent structure or an integrated structure integrated together. When the processor 301 and the memory 302 are in an independent structure, the memory 302 and the processor 301 may be coupled and connected through a bus 303.

[0156] The electronic device in this embodiment may execute the technical solutions in the above method, and the specific implementation process and technical principle are the same, which will not be elaborated here.

[0157] Embodiment 4 also provides a computer-readable storage medium, including a computer program and instructions. When the computer program or instructions run on a computer, the computer is enabled to execute the heavy rain fall area identification method based on the fusion wind field of geostationary satellite cloud-derived winds according to any embodiment of the present invention.

[0158] The computer-readable storage medium includes various media that can store program codes, such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs.

[0159] This embodiment also provides a computer program product. The computer program product includes a computer program. The computer program is stored in a readable storage medium. At least one processor of the electronic device can read the computer program from the readable storage medium, and at least one processor executes the computer program to enable the electronic device to execute the solution provided in any of the above embodiments.

[0160] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recorded in the disclosure of the present invention can be executed in parallel, sequentially, or in different orders,

[0161] as long as the desired results of the technical solution disclosed in the present invention can be achieved, which is not limited herein.

[0162] In summary, the heavy rain fall area identification method and system based on the fusion wind field of geostationary satellite cloud-derived winds of the present invention have the following beneficial effects:

[0163] 1. Compared with the traditional optimal interpolation method, the adaptive intelligent optimal interpolation method adopted by the present invention has significant advantages; the adaptive algorithm can dynamically adjust the interpolation coefficient according to real-time data and environmental changes, improving the accuracy and reliability of the results; by training and optimizing the R matrix through a convolutional neural network, complex non-linear relationships can be captured, providing more accurate estimates; considering the characteristics of satellite cloud-derived winds and model wind field forecast data comprehensively, the intelligent algorithm can make the most of the advantages of various types of data, enhancing the wind field estimation effect; in addition, the adaptive intelligent optimal interpolation method has stronger adaptability, can handle the data fusion requirements of different atmospheric vertical layers, and provide stable and reliable interpolation results under rapidly changing meteorological conditions, thus significantly improving the correction effect of the upper-level wind field and the forecast accuracy of the heavy rain fall area;

[0164] 2. Establishing a three-dimensional and refined fusion wind field product based on geostationary meteorological satellite cloud-derived winds can reflect the characteristics of synoptic-scale and mesoscale weather systems, conform to the usage habits of forecasters, make up for the problem that a single data source cannot well capture the occurrence and development of weather systems, and realize continuous dynamic monitoring of severe convective weather. Satellite cloud-derived winds can not only effectively correct the wind field of the model, but also make up for the problem of relatively low spatio-temporal resolution of the sounding wind field;

[0165] 3. Diagnostic quantities such as vorticity, divergence, and vertical wind shear based on the calculated integrated wind field, which are related to the position of the strong center of convective weather distribution, can effectively monitor and forecast the potential rainfall areas of heavy rain. Forecasters can correct the forecast of heavy rain areas by the model according to this integrated wind field and products, thereby improving the forecasting accuracy of heavy rain areas.

[0166] Note that the above is only a preferred embodiment of the present invention and the technical principles applied. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the concept of the present invention, more other equivalent embodiments can be included, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for identifying heavy rain fall areas based on a fused wind field of geostationary satellite cloud-derived winds, characterized in that, Including: Obtain the original cloud drift wind data of geostationary meteorological satellites, and perform preprocessing to generate high-quality cloud drift wind data of geostationary satellites; Obtain the numerical model forecast wind field in real time, and fuse the high-quality cloud drift wind data of the geostationary satellite and the numerical model forecast wind field to generate three-dimensional fused wind field data; Based on the three-dimensional fused wind field data, calculate and generate diagnostic physical quantities through a first preset method; Based on the diagnostic physical quantities, identify the heavy rain falling area to generate a first heavy rain falling area; Revise the forecast heavy rain falling area based on the first heavy rain falling area to generate a second heavy rain falling area; Among them, the obtaining the numerical model forecast wind field in real time, and fusing the high-quality cloud drift wind data of the geostationary satellite and the numerical model forecast wind field to generate three-dimensional fused wind field data includes: Obtain the numerical model forecast wind field in real time; Fuse the high-quality cloud drift wind data of the geostationary satellite and the numerical model forecast wind field through a second preset method to generate the three-dimensional fused wind field data; Among them, the fusing the high-quality cloud drift wind data of the geostationary satellite and the numerical model forecast wind field through a second preset method to generate the three-dimensional fused wind field data includes: Calculate the background error covariance matrix B of the numerical model forecast wind field; Calculate the observation error covariance matrix R of the high-quality cloud drift wind data of the geostationary satellite; Based on the background error covariance matrix B and the observation error covariance matrix R, fuse the high-quality cloud drift wind data of the geostationary satellite and the numerical model forecast wind field through an adaptive optimal interpolation algorithm to generate the three-dimensional fused wind field data; Among them, the adaptive optimal interpolation algorithm calculates the sounding data and the model background field data at different pressure altitude levels through a statistical method of regression analysis, and calculates the error distribution and statistical variance at each pressure altitude level, so as to be able to estimate the optimal value respectively; Among them, the specific formula for calculating the background error covariance matrix B using the adaptive optimal interpolation algorithm is: ； Among them, represents the sum of squared errors of the j-th grid point, represents the statistical variance of the j-th grid point, represents the parameter of different pressure altitude layers h .

2. The method for identifying heavy rain falling area based on the fusion wind field of geostationary satellite cloud-derived wind as claimed in claim 1, wherein The obtaining the original cloud drift wind data of geostationary meteorological satellites, and performing preprocessing to generate high-quality cloud drift wind data of geostationary satellites includes: Obtain the cloud drift wind information of the water vapor channel and the cloud drift wind information of the infrared channel of the geostationary meteorological satellite; Integrate the cloud drift wind information of the water vapor channel and the cloud drift wind information of the infrared channel to generate the original cloud drift wind data; Perform quality control on the original cloud drift wind data to generate the high-quality cloud drift wind data of the geostationary satellite.

3. A heavy rain fall area identification system based on a wind field fused by geostationary satellite cloud-derived winds, characterized in that, Including: The first generation module is used to obtain the original cloud drift wind data of geostationary meteorological satellites, and perform preprocessing to generate high-quality cloud drift wind data of geostationary satellites; The second generation module is used to obtain the numerical model forecast wind field in real time, and fuse the high-quality cloud drift wind data of the geostationary satellite and the numerical model forecast wind field to generate three-dimensional fused wind field data; The third generation module is used to calculate and generate diagnostic physical quantities based on the three-dimensional fused wind field data through a first preset method; The fourth generation module is used to identify the heavy rain falling area based on the diagnostic physical quantities to generate a first heavy rain falling area; And The fifth generation module is used to revise the forecast heavy rain falling area based on the first heavy rain falling area to generate a second heavy rain falling area; Among them, the second generation module includes: The second obtaining unit is used to obtain the numerical model forecast wind field in real time; A third generation unit, configured to generate the three-dimensional fused wind field data by fusing the high-quality cloud-derived wind data of the geostationary satellite and the numerical model forecast wind field through a second preset method; Wherein, the third generation unit includes: A first calculation subunit, configured to calculate the background error covariance matrix B of the numerical model forecast wind field; A second calculation subunit, configured to calculate the observation error covariance matrix R of the high-quality cloud-derived wind data of the geostationary satellite; A generation subunit, configured to fuse the high-quality cloud-derived wind data of the geostationary satellite and the numerical model forecast wind field based on the background error covariance matrix B and the observation error covariance matrix R through an adaptive optimal interpolation algorithm, to generate the three-dimensional fused wind field data; Among them, the adaptive optimal interpolation algorithm calculates the sounding data and the model background field data at different pressure altitude levels through a statistical method of regression analysis, and calculates the error distribution and statistical variance at each pressure altitude level, so as to be able to estimate the optimal value respectively; Wherein, the specific formula for calculating the background error covariance matrix B by using the adaptive optimal interpolation algorithm is: ； Among them, represents the sum of squared errors of the j-th grid point, represents the statistical variance of the j-th grid point, represents the parameter of different barometric altitude layers h .

4. The heavy rain area identification system based on the fusion wind field of geostationary satellite cloud-derived wind as claimed in claim 3, wherein The first generation module includes: A first acquisition unit, configured to acquire the water vapor channel cloud-derived wind information and the infrared channel cloud-derived wind information of the geostationary meteorological satellite; A first generation unit, configured to integrate the water vapor channel cloud-derived wind information and the infrared channel cloud-derived wind information to generate the original cloud-derived wind data; and A second generation unit, configured to perform quality control on the original cloud-derived wind data to generate the high-quality cloud-derived wind data of the geostationary satellite.

5. An electronic device, characterized in that, Comprising: At least one processor; And A memory communicatively connected to the at least one processor; Wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can execute the heavy rainfall area identification method based on the geostationary satellite cloud-derived wind fused wind field according to any one of claims 1-2.

6. A computer-readable storage medium, characterized in that, Comprising a computer program and instructions, when the computer program or the instructions run on a computer, the computer is caused to execute the heavy rainfall area identification method based on the geostationary satellite cloud-derived wind fused wind field according to any one of claims 1-2.

Images