Typhoon long-distance rainstorm rapid identification method based on AI meteorological model, storage medium and computer program product

Abstract

The invention discloses a typhoon long-distance rainstorm rapid identification method based on an AI meteorological model, a storage medium and a computer program product, and the method comprises the steps: obtaining historical weather data and / or forecast weather data, selecting an AI weather forecast model which is based on a deep learning technology and carries out the training through a large amount of historical weather data, representative and typical rainfall examples are screened, a control experiment and a sensitivity experiment are designed, and typhoon core vortexes are removed from an initial field. By comparing the difference between the results of the control experiment and the sensitivity experiment, the influence of the typhoon on the long-distance rainfall is quantified based on the whole-layer water vapor divergence change and rainfall change analysis, and the long-distance rainstorm influence range and intensity of the typhoon are identified and determined. According to the method, the precision of medium-term weather forecast and extreme weather event recognition is remarkably improved, efficient calculation and wide practical application value are achieved, a scientific basis is provided for disaster early warning and disaster reduction management, and a new method is provided for meteorological scientific research and teaching.

Description

Technical Field

[0001] The present invention belongs to the field of meteorological science and artificial intelligence technology, especially the interdisciplinary field of numerical weather forecasting, disaster warning and disaster reduction management, and the application of artificial intelligence technology in the field of meteorology. Specifically, the present invention focuses on using advanced artificial intelligence models (such as deep learning, machine learning, etc.) to improve the recognition and prediction capabilities of typhoons and their long-distance precipitation effects, and proposes a typhoon long-distance rainstorm rapid identification method based on AI meteorological models, storage medium and computer program products, aiming to provide new technical means for early warning and disaster prevention and reduction of extreme weather events by optimizing computing efficiency and prediction accuracy. Background Art

[0002] Tropical cyclones (TCs) are one of the most destructive natural disasters, causing heavy rainfall, floods and landslides in affected areas. Although these extreme precipitation events are mainly concentrated in the direct impact area of ​​the TC, their long-range effects can also have catastrophic impacts. This long-range effect is often referred to as "typhoon long-range heavy rain", which is often obscured by the precipitation directly caused by the TC in research. Understanding and identifying the long-range impact of typhoons on precipitation is an important aspect of meteorological research, which is crucial to improving the early warning capabilities of meteorological disasters, disaster prevention and mitigation, and supporting effective weather decision-making processes.

[0003] Traditional numerical weather prediction models (NWP), such as the Weather Research and Forecasting Model (WRF), have been widely used to study typhoons and related phenomena. These models help meteorologists understand the physical mechanisms of typhoons through simulations and sensitivity experiments. However, these models often require high computational costs to handle complex atmospheric dynamics and thermodynamic processes, limiting their efficiency and wide application in practical applications.

[0004] With the advancement of artificial intelligence technology, innovative methods such as machine learning and deep learning technology have been introduced into the field of weather forecasting. This transformative method provides higher prediction accuracy and computational efficiency than traditional models. Some innovative AI systems, such as Pangu-Weather (hereinafter referred to as Pangu), GraphCast, and Fengwu, have become leading models for medium-term global weather forecasts. As open source platforms, these AI models have promoted cooperation and improvement in the global research community and have been adopted by multiple international forecast centers. AI models have significant advantages in processing large amounts of data and complex pattern recognition, making them increasingly widely used in weather forecasting.

[0005] However, despite the excellent performance of AI models in weather forecasting, there are still doubts about whether they can accurately capture and describe complex physical processes, especially the impact of typhoons on weather patterns. Understanding the long-range effects of typhoons usually requires sensitivity experiments to fully analyze their physical mechanisms, which is computationally expensive and inefficient for traditional numerical models. Therefore, how to use advanced artificial intelligence models (such as deep learning, machine learning, etc.) to improve the recognition and prediction capabilities of tropical cyclones and their long-range precipitation effects, improve the recognition efficiency and accuracy of typhoon long-range rainstorms, and thus improve the prediction accuracy and computational efficiency of extreme precipitation events, is a technical problem that needs to be solved urgently. Summary of the invention

[0006] 1. Purpose of the invention In order to solve at least one of the above-mentioned and other technical problems in the prior art, the present invention provides a method for quickly identifying typhoon long-distance heavy rain based on an AI meteorological model, a storage medium and a computer program product, aiming to improve the recognition accuracy and efficiency of the long-distance impact of typhoons. The method uses weather data and combines it with an AI forecast model to design control experiments and sensitivity experiments to analyze the long-distance impact of typhoons on precipitation. The sensitivity experiment removes the typhoon core vortex, thereby realizing the rapid identification of typhoon long-distance heavy rain. At the same time, by comparing the prediction results of the control experiment and the sensitivity experiment, the prediction accuracy and calculation efficiency of extreme precipitation events are significantly improved, which is of great significance to disaster warning and disaster reduction management.

[0007] (II) Technical solution In order to achieve the purpose of the invention and solve the technical problems, the present invention adopts the following technical solutions: The first invention object of the present invention is to provide a method for quickly identifying typhoon long-distance rainstorms based on an AI meteorological model, so as to improve the recognition accuracy and efficiency of the long-distance impact of typhoons. Specifically, the method for quickly identifying typhoon long-distance rainstorms includes at least the following steps when implemented: SS1. Weather data acquisition Acquire high-quality and high-temporal and high-spatial resolution historical weather data and / or forecast weather data, which covers global land and ocean areas and includes at least temperature, humidity, air pressure, wind speed, wind direction and precipitation meteorological parameters to ensure high accuracy and reliability of model initial conditions; SS2. AI weather forecast model selection Select an AI weather model based on deep learning technology and trained with a large amount of historical weather data, which can provide high-precision medium-term weather forecasts and identify extreme weather events as the forecast model. The selected model must be capable of processing complex meteorological data and quickly generating high-precision forecast results. SS3. Selection of typical precipitation cases Representative and typical precipitation cases should be screened and determined based on historical meteorological disaster records and historical meteorological data. The selected typical precipitation cases should have certain characteristics of typhoon long-distance impact, and at least the reporting time, precipitation location information and typhoon location information of the typical precipitation cases should be provided; SS4. Control and sensitivity test design Based on the selected AI weather forecast model, the acquired weather data and the selected typical precipitation cases, a control experiment (CTRL) and a sensitivity experiment (RMTC) are designed, where: The control experiment is based on the selected AI weather forecast model, using historical weather data and / or forecast weather data corresponding to the selected typical precipitation case as the initial field, performing standard weather forecast simulation and generating benchmark prediction results; The sensitivity experiment, based on the control experiment, applies the method of removing the typhoon core vortex from the initial field to construct a new initial field, and re-simulates the weather forecast based on the selected AI weather forecast model to generate sensitivity experiment results; SS5. Result analysis and identification of long-distance typhoon heavy rain By comparing the differences between the results of control experiments and sensitivity experiments, and based on the analysis of changes in water vapor divergence and / or precipitation changes in the entire layer, the impact of the target typhoon on long-distance precipitation is quantitatively analyzed, and the impact range and intensity of the typhoon's long-distance rainstorm are identified and determined, thereby achieving rapid identification of the typhoon's long-distance rainstorm.

[0008] The second invention object of the present invention is to provide a computer program product, including computer instructions, for executing the above-mentioned typhoon long-distance heavy rain rapid identification method based on AI meteorological model.

[0009] The third invention object of the present invention is to provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned typhoon long-distance rainstorm rapid identification method based on AI meteorological model is implemented.

[0010] (III) Technical Effect Compared with the prior art, the method, storage medium and computer program product for quickly identifying typhoon long-distance rainstorm based on AI meteorological model of the present invention have the following beneficial and significant technical effects: (1) The present invention uses an AI weather forecast model, which can provide more accurate forecasts than traditional numerical weather forecast models by using advanced deep learning algorithms and a large amount of historical meteorological data for training. For example, the Pangu model performs particularly well in medium-term weather forecasts and extreme weather event identification, significantly improving the accuracy of weather forecasts. In addition, by using high-resolution reanalysis data and historical weather data or forecast weather data, the high accuracy and reliability of the model's initial conditions are ensured, providing a solid data foundation for the forecast results.

[0011] (2) The present invention has significant advantages in computational efficiency. Compared with traditional numerical weather forecast models, AI weather forecast models can process and analyze large-scale data in a shorter time, thereby providing more timely forecast results. By applying the method of removing the typhoon core vortex, the system of the present invention can effectively conduct sensitivity experiments, reduce computational complexity, and improve computational efficiency. This method can accurately remove the typhoon core vortex from the initial field, thereby better analyzing the long-range impact of the typhoon and enhancing the ability to identify and forecast extreme weather events.

[0012] (3) The present invention has important practical application value in disaster warning and disaster reduction management. By improving the recognition accuracy and efficiency of typhoon long-distance rainstorms, it provides a scientific basis for disaster warning and disaster reduction management, helping decision makers to formulate more effective response strategies. The system uses refined analysis of the whole-layer water vapor flux and the whole-layer water vapor divergence to accurately capture the long-distance impact of typhoons, thereby significantly improving the accuracy and reliability of forecasts. This is of great significance for timely warning and taking disaster prevention measures.

[0013] (4) The present invention also has broad application prospects in meteorological research and teaching. Its innovative methods and refined analysis tools provide new tools and methods for meteorological research, which helps to deeply understand the long-range impact mechanism of typhoons and explore the impact of weather systems of different scales on precipitation. By comparing control experiments and sensitivity experiments, the system can quantitatively analyze the specific impact of the typhoon core vortex on long-range precipitation, and combine wavelength filtering analysis to explore the impact of weather systems of different scales on precipitation.

[0014] (5) The present invention improves the accuracy and efficiency of long-distance typhoon rainstorm identification by using advanced AI technology and high-resolution meteorological data, significantly enhances the forecasting capability and computing efficiency, and has significant technical effects and broad application prospects in meteorological forecasting, disaster warning and disaster reduction management, as well as meteorological scientific research and teaching. Through the above advantages and innovations, the present invention provides strong technical support for meteorological forecasting and disaster warning, and provides a scientific solution for dealing with extreme weather events. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The figure shows a schematic diagram of the implementation process of the method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model of the present invention; Figure 2 The figure shows a schematic diagram of the architecture of the typhoon long-distance rainstorm rapid identification system based on the AI ​​meteorological large model of the present invention; Figure 3 The figure shows the time series of the whole-layer water vapor flux divergence averaged over a certain region of East Asia from 12:00 UTC on July 28, 2023 to 00:00 UTC on August 1, 2023 (10 4 ·g·m -2 ·s -1 ) schematic diagram showing the comparison results between Pangu and Pangu_RMTC. DETAILED DESCRIPTION

[0016] In order to better understand the present invention, the content of the present invention is further explained in conjunction with the embodiments below. In the accompanying drawings, the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be understood as limitations on the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. The structure and technical solution of the present invention are further described in detail below in conjunction with the accompanying drawings, and an embodiment of the present invention is given.

[0017] The present invention aims to provide a method, storage medium and computer program product for quickly identifying typhoon long-distance rainstorms based on an AI meteorological model, so as to improve the recognition accuracy and efficiency of the long-distance impact of typhoons. The method is based on weather data and combined with an AI meteorological forecast model to design control experiments and sensitivity experiments to analyze the long-distance impact of typhoons on precipitation. The sensitivity experiment removes the typhoon core vortex to achieve rapid identification of typhoon long-distance rainstorms. At the same time, the method significantly improves the prediction accuracy and calculation efficiency of extreme precipitation events by comparing the prediction results of the control experiment and the sensitivity experiment, which is of great significance to disaster warning and disaster reduction management.

[0018] As a specific example, Figure 1 , 2 As shown, the method for quickly identifying typhoon long-distance rainstorm based on AI meteorological model provided by the present invention mainly includes the following steps when implemented: SS1. Weather data acquisition Acquire high-quality and high-temporal and high-spatial resolution historical weather data and / or forecast weather data, which covers land and ocean areas around the world and includes at least temperature, humidity, air pressure, wind speed, wind direction and precipitation meteorological parameters to ensure high accuracy and reliability of the model initial conditions.

[0019] In some preferred embodiments, when this step is executed, the historical weather data adopts high-resolution global meteorological data provided by the ERA5 weather data set of the European Centre for Medium-Range Weather Forecasts (ECMWF), and the forecast weather data adopts forecast field data provided by the Integrated Forecasting System (IFS) of the European Centre for Medium-Range Weather Forecasts (ECMWF).

[0020] The ERA5 weather dataset is the fifth generation of global climate reanalysis dataset provided by the European Centre for Medium-Range Weather Forecasts (ECMWF), covering high-resolution weather data from 1979 to the present. It provides long-term series of global climate data by combining global observational data and numerical weather forecast models, and is widely used in meteorological research, climate change analysis and environmental monitoring. ERA5 data includes a variety of meteorological variables, such as temperature, humidity, wind speed and direction, precipitation, air pressure, etc., covering land and ocean areas around the world. The data resolution of ERA5 is 0.25°×0.25° (about 31 kilometers), and there are 37 pressure layers in the vertical direction, from the ground to the top of the atmosphere. ERA5 provides hourly data, which is helpful for analyzing short-term weather changes and extreme weather events. The ERA5 dataset integrates satellite observations, ground observations and weather forecast model outputs, and is generated by four-dimensional variational data assimilation (4D-Var) technology.

[0021] ECMWF's Integrated Forecasting System (IFS) is one of the world's leading numerical weather prediction models, providing high-precision medium-term weather forecast data. The IFS forecast field generates weather forecasts for the next few days to weeks through complex numerical models and observational data assimilation. The IFS forecast field includes various meteorological parameters such as temperature, humidity, wind speed and direction, precipitation, air pressure, etc., which is similar to ERA5, but focuses on future weather forecasts. The IFS data resolution can reach about 9 kilometers (T1279), and there are 137 levels in the vertical direction, from the ground to the top of the atmosphere at 0.01 hPa. IFS forecast data usually provides hourly or three-hour forecast outputs, and the time span can range from several days to weeks. The IFS forecast system combines global observation data (including satellite, radar and ground observations) and uses advanced data assimilation technology and high-performance computing platforms to generate forecasts.

[0022] SS2. AI weather forecast model selection An AI weather model based on deep learning technology and trained with a large amount of historical weather data, which is capable of providing high-precision medium-term weather forecasts and identification of extreme weather events, is selected as the forecasting model. The selected model has the ability to process complex meteorological data and quickly generate high-precision forecast results.

[0023] In some preferred examples, when step SS2 is implemented, the AI ​​weather forecast model selects Pangu, GraphCast, Fengwu or other high-precision AI models with similar performance, and combines multimodal data input, including meteorological observation data, historical weather forecast data and environmental data, to enhance the model's predictive ability, and improve the recognition and forecast accuracy of complex weather events through cross-modal learning technology. Specifically, meteorological observation data include but are not limited to ground observation station data, meteorological satellite remote sensing data and radar observation data; historical weather forecast data include global and regional weather forecast data for the past few years; environmental data include key parameters such as ocean temperature, humidity, wind speed and air pressure. Through cross-modal learning technology, the model can extract richer feature information from multimodal data, further improve the recognition and forecast accuracy of complex weather events, especially in the face of the interactive influence of multiple meteorological systems, and can more accurately predict the long-range impact of typhoons and the occurrence of heavy rain events.

[0024] SS3. Selection of typical precipitation cases Representative and typical precipitation cases are screened and determined based on historical meteorological disaster records and historical meteorological data. The selected typical precipitation cases should have certain characteristics of typhoon long-distance impact, and at least provide the starting time, precipitation location information and typhoon location information of the typical precipitation case.

[0025] Specifically, in this step SS3, the screening criteria for typical precipitation cases include precipitation intensity exceeding a preset critical threshold, precipitation duration exceeding a preset critical value, and / or the precipitation event having the characteristics of being caused by a tropical cyclone or other extreme weather system. By clustering historical meteorological disaster records and historical meteorological data, several representative typical precipitation cases are determined, providing research objects for subsequent sensitivity experimental studies.

[0026] SS4. Control and sensitivity test design Based on the selected AI weather forecast model, the acquired weather data and the selected typical precipitation cases, a control experiment (CTRL) and a sensitivity experiment (RMTC) are designed, where: The control experiment is based on the selected AI weather forecast model, uses the historical weather data and / or forecast weather data corresponding to the selected typical precipitation cases as the initial field, performs standard weather forecast simulation and generates benchmark prediction results; the sensitivity experiment, based on the control experiment, applies the method of removing the typhoon core vortex from the initial field to construct a new initial field, and re-performs the weather forecast simulation based on the selected AI weather forecast model to generate sensitivity experiment results.

[0027] In some preferred embodiments, the Kurihara method is used in the sensitivity experiment to remove the typhoon core vortex from the initial field to construct a new initial field. The method effectively separates and removes the typhoon core vortex in the initial field by applying a cylindrical filter, so that the sensitivity experiment can better analyze the long-range impact of the typhoon, and specifically includes the following sub-steps: SS41. Determine the core location of the typhoon Based on the typhoon's location information and at least in combination with the initial field's air pressure and wind field meteorological data, the location of the typhoon's core is determined, and the typhoon's center is located by analyzing the lowest point of air pressure and the area with the highest wind speed; SS42. Define filter area A cylindrical filtering area is defined around the core position of the typhoon, wherein the radius and height of the filtering area are determined according to the actual size and impact range of the typhoon to cover the entire typhoon vortex structure; SS43. Extracting typhoon vortex A cylindrical filter is used to extract the typhoon vortex from the initial field by calculating the average value in the filter area and separating the symmetrical part of the typhoon through the average value. SS44. Remove the vortex structure By replacing the vortex part in the filter area with the average meteorological parameter value of the surrounding area, the influence of the typhoon core on the initial field is eliminated, thereby removing the extracted typhoon vortex structure from the initial field; SS45. Constructing a new initial field The initial conditions of the numerical weather forecast model are reconstructed using the initial field after removing the typhoon core vortex. The new initial field is used to conduct sensitivity experiments to analyze the long-range impact of the typhoon.

[0028] Through the above sub-steps, the typhoon core vortex in the initial field can be accurately removed, ensuring that the sensitivity experiment can more accurately analyze the impact mechanism of typhoons on long-distance heavy rains.

[0029] In a further preferred embodiment, in the above sub-step SS44, the extracted typhoon vortex structure is removed from the initial field by the following steps: SS441. Assume that the meteorological variable in the initial field is Φ(x , y , z ), the typhoon core position is ( x 0, y 0), the radius of the filtering area is R , then the cylindrical filter can be expressed as: SS442. Calculate the typhoon vortex component of various meteorological variables (such as air pressure, wind speed, wind direction, etc.) in the initial field based on the following calculation formula: Among them, Φ b ( x , y , z ) represents the background field, which can usually be calculated by filtering: SS443. Subtract the vortex component calculated in the above substep SS442 from the original initial field to generate an initial condition without the typhoon core vortex: In the formula, Φ new ( x , y , z ) represents the meteorological variables in the new initial field without the typhoon core vortex, Φ( x , y , z ) is the meteorological variable in the original initial field, Φ v ( x , y , z ) is the typhoon vortex component of each meteorological variable.

[0030] SS5. Result analysis and identification of long-distance typhoon heavy rain By comparing the differences between the results of control experiments and sensitivity experiments, and based on the analysis of changes in water vapor divergence and / or precipitation changes in the entire layer, the impact of the target typhoon on long-distance precipitation is quantitatively analyzed, and the impact range and intensity of the typhoon's long-distance rainstorm are identified and determined, thereby achieving rapid identification of the typhoon's long-distance rainstorm.

[0031] In some preferred embodiments, the whole-layer water vapor flux and the whole-layer water vapor divergence are calculated in the following manner in this step, wherein: (a) Calculation of the whole-layer water vapor flux. The whole-layer water vapor flux is used to indicate the amount of water vapor transported in the atmosphere under the action of the horizontal wind field. The calculation formula is as follows: in, qis the specific humidity, V is the horizontal wind speed, V= ( u , v ), are the east-west wind speed component and the north-south wind speed component, p s , p t are the air pressure at the surface and at the top of the atmosphere, respectively; (b) Calculation of water vapor divergence of the entire layer. The water vapor divergence of the entire layer indicates the convergence or divergence of water vapor in the atmosphere. It is the divergence of water vapor flux and its calculation formula is as follows: Where Q is the water vapor flux of the entire layer, Q u , Q v are the east-west and north-south components of water vapor flux, Represents the divergence of water vapor flux.

[0032] In step SS4, both the control experiment (CTRL) and the sensitivity experiment (RMTC) are based on the selected AI weather forecast model, using historical weather data and / or forecast weather data corresponding to typical precipitation cases as the initial field. The control experiment uses the complete initial field data to perform standard weather forecast simulations to generate benchmark forecast results. Based on the control experiment, the sensitivity experiment applies the Kurihara method to remove the typhoon core vortex from the initial field, and effectively separates and removes the typhoon core vortex in the initial field through a cylindrical filter, thereby constructing a new initial field, and re-performing weather forecast simulations based on this new initial field to generate sensitivity experiment results.

[0033] By using Python for drawing, drawing libraries such as Matplotlib and Seaborn can be used to visualize the forecast results. The drawing content may include the entire layer of water vapor, the change in the divergence of the entire layer of water vapor, the distribution of precipitation and its evolution over time, etc., so as to facilitate a more intuitive understanding of the long-distance impact of typhoons.

[0034] Figure 3 The time series of the whole-layer water vapor flux divergence averaged over a certain region of East Asia from 12:00 UTC on July 28, 2023 to 00:00 UTC on August 1, 2023 (10 4 ·g·m -2 ·s -1 ), from the results of Pangu and Pangu_RMTC. The figure shows the comparison between Pangu and Pangu_RMTC, which shows the trend of the change of the water vapor flux divergence in the whole layer after the typhoon core vortex is removed. These charts provide an important reference for understanding the impact mechanism of typhoon long-distance heavy rain through quantitative and visual analysis.

[0035] SS6. Evaluation and verification of typhoon long-distance rainstorm identification results The typhoon long-distance rainstorm identification results obtained in step SS5 are compared with the actual observation data to evaluate the accuracy and reliability of the model prediction, and the parameters and algorithms of the AI ​​weather forecast model are optimized and adjusted according to the evaluation results to improve its prediction accuracy and efficiency.

[0036] The above steps constitute a complete typhoon long-distance rainstorm rapid identification system architecture based on AI meteorological large model (such as Figure 2 As shown in the figure, this system can significantly improve the accuracy and efficiency of identifying the long-distance impact of typhoons, thereby providing strong technical support for weather forecasting and disaster warning.

[0037] Based on the above embodiments, the present invention improves the accuracy of medium-term weather forecasts and extreme weather event identification by using the Pangu weather model, high-resolution ERA5 reanalysis data, and ECMWF's IFS forecast data. The system uses a refined analysis of the whole-layer water vapor flux and the whole-layer water vapor divergence to accurately capture the long-range impact of the typhoon, significantly improving the accuracy and reliability of the forecast. Compared with the traditional numerical weather forecast model, the AI ​​model has significant advantages in computational efficiency and can process and analyze large-scale data more timely. The Kurihara method is used to remove the typhoon core vortex, which reduces the computational complexity and improves the computational efficiency of the sensitivity experiment. By comparing the control experiment with the sensitivity experiment, the system can quantitatively analyze the specific impact of the typhoon core vortex on long-range precipitation and deeply understand the long-range impact mechanism of the typhoon. Combined with wavelength filtering analysis, the impact of weather systems of different scales on precipitation is explored. The system has important application value in disaster warning and disaster reduction management, provides a scientific basis for decision makers, and provides new tools and methods for meteorological research and teaching.

[0038] Through the above embodiments, the purpose of the present invention is fully and effectively achieved. Those skilled in the art can understand that the present invention includes but is not limited to the contents described in the drawings and the above specific embodiments. Although the present invention has been described with respect to the most practical and preferred embodiments currently considered, it should be understood that the present invention is not limited to the disclosed embodiments, and any modification that does not deviate from the functional and structural principles of the present invention will be included in the scope of the claims.

Claims

1. A method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model, characterized in that: The method for quickly identifying typhoon long-distance rainstorms comprises at least the following steps when implemented: SS1. Weather data acquisition Acquire high-quality and high-temporal and high-spatial resolution historical weather data and / or forecast weather data, which covers global land and ocean areas and includes at least meteorological parameters such as temperature, humidity, air pressure, wind speed, wind direction and precipitation; SS2.AI weather forecast model selection Select an AI weather model based on deep learning technology and trained with a large amount of historical weather data, which can provide high-precision medium-term weather forecasts and identify extreme weather events as the forecast model; SS3. Typical precipitation case selection Based on historical meteorological disaster records and historical meteorological data, representative and typical precipitation cases are screened and determined. The selected typical precipitation cases should have certain characteristics of typhoon long-distance impact, and at least provide the starting time, precipitation location information and typhoon location information of the typical precipitation case; SS4. Control and sensitivity test design Based on the selected AI weather forecast model, the acquired weather data and the selected typical precipitation cases, control experiments and sensitivity experiments are designed, including: The control experiment is based on the selected AI weather forecast model, uses historical weather data and / or forecast weather data corresponding to the selected typical precipitation case as the initial field, performs standard weather forecast simulation and generates benchmark prediction results; The sensitivity experiment, based on the control experiment, applies the method of removing the typhoon core vortex from the initial field to construct a new initial field, and re-simulates the weather forecast based on the selected AI weather forecast model to generate sensitivity experiment results; SS5. Result analysis and identification of long-distance typhoon heavy rain By comparing the differences between the results of control experiments and sensitivity experiments, and based on the analysis of changes in water vapor divergence and / or precipitation changes in the entire layer, the impact of the target typhoon on long-distance precipitation is quantitatively analyzed, and the impact range and intensity of the typhoon's long-distance rainstorm are identified and determined, thereby achieving rapid identification of the typhoon's long-distance rainstorm.

2. The method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model according to claim 1 is characterized in that: In the above step SS1, the historical weather data adopts high-resolution global meteorological data provided by the ERA5 weather data set, and the forecast weather data adopts IFS forecast field data.

3. The method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model according to claim 1 is characterized in that: In the above step SS2, the AI ​​weather forecast model selects Pangu, GraphCast, Fengwu or other similar AI forecast models, and combines multimodal data input, including meteorological observation data, historical weather forecast data and environmental data, to enhance the model's prediction ability, and improve the recognition and forecasting accuracy of complex weather events through cross-modal learning technology.

4. The method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model according to claim 1 is characterized in that: In the above step SS3, the screening criteria for typical precipitation cases include precipitation intensity exceeding a preset critical threshold, precipitation duration exceeding a preset critical value, and / or the precipitation event having the characteristics of being caused by a tropical cyclone or other extreme weather system. By clustering the historical meteorological disaster records and historical meteorological data, several representative typical precipitation cases are determined to provide research objects for subsequent sensitivity experimental studies.

5. The method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model according to claim 1 is characterized in that: In the above step SS4, a new initial field is constructed by removing the typhoon core vortex from the initial field using the Kurihara method, and specifically includes the following sub-steps: SS41. Determine the core location of the typhoon Based on the typhoon's location information and at least in combination with the initial field's air pressure and wind field meteorological data, the location of the typhoon's core is determined, and the typhoon's center is located by analyzing the lowest point of air pressure and the area with the highest wind speed; SS42. Define filter area A cylindrical filtering area is defined around the core position of the typhoon, wherein the radius and height of the filtering area are determined according to the actual size and impact range of the typhoon to cover the entire typhoon vortex structure; SS43. Extracting typhoon vortex A cylindrical filter is used to extract the typhoon vortex from the initial field by calculating the average value in the filter area and separating the symmetrical part of the typhoon through the average value. SS44. Remove the vortex structure By replacing the vortex part in the filter area with the average meteorological parameter value of the surrounding area, the influence of the typhoon core on the initial field is eliminated, thereby removing the extracted typhoon vortex structure from the initial field; SS45. Constructing a new initial field The initial conditions of the numerical weather forecast model are reconstructed using the initial field after removing the typhoon core vortex. The new initial field is used to conduct sensitivity experiments to analyze the long-range impact of the typhoon.

6. The method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model according to claim 5 is characterized in that: In the above sub-step SS44, the extracted typhoon vortex structure is removed from the initial field by: SS441. Assume that the meteorological variables in the initial field are Φ(x,y,z), the typhoon core position is (x0,y0), and the radius of the filter area is R, then the cylindrical filter is expressed as: SS442. Calculate the typhoon vortex component Φ of various meteorological variables including air pressure, wind speed and wind direction in the initial field based on the following calculation formula: v (x,y,z): Φ v (x,y,z)=Φ(x,y,z)-Φ b (x,y,z) Among them, Φ b (x, y, z) represents the background field and is calculated by the filter shown below: SS443. Subtract the vortex components obtained by the technique in substep SS442 from the initial field to generate initial conditions without the typhoon core vortex: Φ new (x,y,z)=Φ(x,y,z)-Φ v (x,y,z) In the formula, Φ new (x, y, z) represents the meteorological variables in the new initial field without the typhoon core vortex, Φ(x, y, z) represents the meteorological variables in the original initial field, Φ v (x, y, z) are the typhoon vortex components of each meteorological variable.

7. The method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model according to claim 1 is characterized in that: In the above step SS5, the whole-layer water vapor flux and the whole-layer water vapor divergence are calculated based on the following method, wherein: (a) Calculation of water vapor flux in the whole layer: The whole-layer water vapor flux is used to indicate the amount of water vapor transported in the atmosphere under the action of the horizontal wind field. The calculation formula is as follows: Where q is the relative humidity, V is the horizontal wind speed, V = (u, v), which are the east-west wind speed component and the north-south wind speed component, respectively, and p s 、p t are the air pressure at the surface and at the top of the atmosphere, respectively; (b) Calculation of water vapor dispersion in the whole layer: The whole layer water vapor divergence indicates the convergence or divergence of water vapor in the atmosphere. It is the divergence of water vapor flux and its calculation formula is as follows: Where Q is the water vapor flux of the whole layer, Q u , Q v are the east-west and north-south components of water vapor flux, Represents the divergence of water vapor flux.

8. The method for rapid identification of typhoon long-distance rainstorm based on AI meteorological model according to claim 1 is characterized in that: The method for quickly identifying typhoon long-distance rainstorms also includes: SS6. Evaluation and verification of typhoon long-distance rainstorm identification results The typhoon long-distance rainstorm identification results obtained in step SS5 are compared with the actual observation data to evaluate the accuracy and reliability of the model prediction, and the parameters and algorithms of the AI ​​weather forecast model are optimized and adjusted according to the evaluation results to improve its prediction accuracy and efficiency.

9. A computer program product, characterized in that It includes computer instructions for executing the method for quickly identifying typhoon long-distance heavy rain based on the AI ​​meteorological model as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the method for quickly identifying typhoon long-distance heavy rain based on the AI ​​meteorological model as described in any one of claims 1 to 8 is implemented.

Images