Method for predicting meteorological disasters based on weather classification

Abstract

The invention discloses a method for predicting meteorological disasters based on weather classification, and relates to the technical field of weather prediction. Acquiring spatial meteorological data of the target area, and performing three-dimensional processing on the spatial meteorological data to obtain training data; obtaining a terrain distribution diagram of the target area, and inputting the training data and the terrain distribution diagram into a weather prediction model for training to obtain a weather prediction result; and determining a weather type of the target area according to the weather prediction result, and performing meteorological disaster prediction according to the weather type. According to the method, the spatial meteorological data of the target area are comprehensively obtained, three-dimensional processing is carried out, and the training data are input into an advanced weather prediction model in combination with detailed terrain distribution map information, so that the prediction capability of the model can be remarkably improved. According to the process, the weather type of the target area can be more accurately determined, powerful support is provided for meteorological disaster prediction, and the prediction accuracy is effectively enhanced.

Description

Technical Field

[0001] The invention belongs to the technical field of weather forecasting, and in particular relates to a method for forecasting meteorological disasters based on weather classification. Background Art

[0002] Meteorological disaster prediction based on weather classification is a method that analyzes the characteristics of different weather types and their correlation with meteorological disasters, and uses historical meteorological data and advanced algorithms to establish a prediction model to predict possible future meteorological disasters and their intensity. This method provides strong support for meteorological disaster warning, disaster prevention and mitigation, etc. by applying the model to current weather data in real time. With the improvement of meteorological observation technology and computing power, it is continuously optimized and developed to further improve the prediction accuracy and timeliness, and provide a more reliable reference for the prevention and response to meteorological disasters.

[0003] Patent CN118427639A, the present invention provides a method and device for generating meteorological forecast results based on multi-dimensional meteorological data; obtain the meteorological observation data sequence of each meteorological station in the target area, on this basis, analyze the change trend index of each meteorological observation data sequence, and mine the inherent laws and correlations between the data of multiple meteorological stations according to the change trend index, and combine it with the location information of the meteorological station to finally generate the meteorological forecast result. Thus, the accuracy of the meteorological forecast result is improved.

[0004] In the above technologies, the weather forecast data is not updated in a timely manner and the weather forecast data is single, resulting in inaccurate prediction results. Summary of the invention

[0005] The purpose of the present invention is to solve the problems of untimely data update and single meteorological forecast data in existing meteorological forecasts, resulting in inaccurate forecast results, and to propose a method for predicting meteorological disasters based on weather classification.

[0006] In a first aspect of the present invention, a method for predicting meteorological disasters based on weather typing is first proposed, the method comprising: Acquire space weather data of the target area, and perform three-dimensional processing on the space weather data to obtain training data; the space weather data is data collected at different altitudes in the target area; Obtaining a terrain distribution map of a target area, and inputting the training data and the terrain distribution map into a weather prediction model for training to obtain a weather prediction result; The weather type of the target area is determined according to the weather forecast result, and meteorological disaster prediction is performed according to the weather type.

[0007] Optionally, the space weather data includes: ground weather station data, high-altitude weather observation data and satellite observation data, and the space weather data is processed three-dimensionally to obtain training data including: The satellite observation data is divided into a target sub-region set according to a preset interval, and the ground meteorological station data and the high-altitude meteorological observation data are spatially interpolated for any target sub-region according to a preset height to obtain a spatial distribution map; Arranging the spatial distribution graphs in time sequence to obtain a time series graph group; the time series graph group includes the spatial distribution graph at any time; Meteorological attribute information is added to the time series graph group to obtain training data; the meteorological attribute information includes: air temperature, air humidity, wind direction and wind speed.

[0008] Optionally, before inputting the training data and the terrain distribution map into a weather prediction model for training to obtain a weather prediction result, the method includes: Acquire historical training data and historical terrain distribution maps of the target area, and input the historical training data and the historical terrain distribution maps into the machine learning model for training in a time series; The output of the machine learning model is used as a target model parameter, and the parameters of the fully connected layer, the convolutional layer, and the recurrent layer of the machine learning model are updated; The machine learning model with updated parameters is used as a weather prediction model; the output of the weather prediction model is the training data of the target sub-area in the future time period.

[0009] Optionally, before inputting the training data into the weather prediction model for training to obtain the weather prediction result, the following steps are performed: Performing update status detection on the training data, detecting the timestamp of data update and determining the data update frequency, and calculating the data update coefficient; Performing data integrity check on the training data, comparing the weather classification with historical data to determine whether there is data missing, and calculating the data missing coefficient; Performing outlier detection on the training data, determining whether tampering has occurred, and calculating a data tampering coefficient; A data evaluation coefficient is calculated according to the data update coefficient, the data missing coefficient and the data tampering coefficient, and the training data is evaluated according to the data evaluation coefficient.

[0010] Optionally, the data evaluation coefficient calculated according to the data update coefficient, the data missing coefficient and the data tampering coefficient includes: Calculation formula of coefficient through data evaluation The data evaluation coefficient is calculated; Among them, F is the data evaluation score, G is the data update coefficient, C is the data tampering coefficient, Q is the data missing coefficient, α and β are constant proportional coefficients, and both α and β are not zero.

[0011] Optionally, evaluating the training data according to the data evaluation coefficient includes: If the data evaluation coefficient is less than the data threshold, an alarm is issued and space weather data is collected again; If the data evaluation coefficient is not less than the data threshold, the training data is input into the weather prediction model for prediction.

[0012] Optionally, forecasting meteorological disasters according to the weather classification includes: Fitting the wind directions in each weather forecast result to obtain a wind direction curve of the target area, and calculating the cloud displacement in the satellite observation data according to the wind speed; Determine the cloud accumulation area of ​​the target area according to the cloud displacement in the satellite observation data, and determine the weather type of the cloud accumulation area according to the ground meteorological station data and the high-altitude meteorological observation data in the weather forecast result; The type of meteorological disaster is determined according to the weather classification. If the impact range of the meteorological disaster type is greater than the area threshold and the intensity of the meteorological disaster type is greater than the intensity threshold, a meteorological disaster warning is issued according to the meteorological disaster type.

[0013] Beneficial effects of the present invention: The present invention proposes a method for predicting meteorological disasters based on weather classification, by acquiring spatial meteorological data of a target area, performing three-dimensional processing on the spatial meteorological data to obtain training data; acquiring a topographic distribution map of the target area, inputting the training data into a weather prediction model for training to obtain a weather prediction result; determining the weather classification of the target area according to the weather prediction result, and predicting meteorological disasters according to the weather classification. The spatial meteorological data of the target area is comprehensively acquired and three-dimensionally processed, combined with detailed topographic distribution map information, and these training data are input into an advanced weather prediction model, which can significantly improve the prediction ability of the model. This process can more accurately determine the weather classification of the target area, thereby providing data support for meteorological disaster prediction and effectively enhancing the accuracy of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below in conjunction with the accompanying drawings.

[0015] Figure 1 A flowchart of a method for predicting meteorological disasters based on weather classification is provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the present invention, the description of "first", "second", etc. is only used for descriptive purposes, and cannot be understood as indicating or implying its relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0017] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0018] The embodiment of the present invention provides a method for predicting meteorological disasters based on weather classification. Figure 1 , Figure 1 A flow chart of a method for predicting meteorological disasters based on weather classification provided by an embodiment of the present invention. The method comprises the following steps: S101, obtaining space weather data of a target area, and performing three-dimensional processing on the space weather data to obtain training data; S102, obtaining a terrain distribution map of the target area, and inputting the training data and the terrain distribution map into a weather prediction model for training to obtain a weather prediction result; S103, determining the weather type of the target area according to the weather forecast result, and performing meteorological disaster prediction according to the weather type.

[0019] Space weather data is data collected at different altitudes in the target area; According to a method for predicting meteorological disasters based on weather classification provided by an embodiment of the present invention, by comprehensively acquiring spatial meteorological data of the target area and performing three-dimensional processing, combined with detailed terrain distribution map information, these training data are input into an advanced weather prediction model, which can significantly improve the prediction ability of the model. This process can more accurately determine the weather classification of the target area, thereby providing data support for meteorological disaster prediction and effectively enhancing the accuracy of the prediction.

[0020] In one implementation, the space weather data is data collected at different altitudes in the target area, that is, monitoring data at three altitudes: ground weather station data, high-altitude weather observation data, and satellite observation data. The altitudes are arranged from high to low, with satellite observation data > high-altitude weather observation data > ground weather station data. Satellite observation data is collected by meteorological satellites, high-altitude weather observation data is collected by data detectors carried by weather balloons, and ground weather station data is collected by ground facilities, such as thermometers, hygrometers, barometers, anemometers, wind vanes, rain gauges, etc.

[0021] In one implementation, the training data is processed in three dimensions, and the weather data is processed in three dimensions and the model is trained. The resulting three-dimensional meteorological model can more intuitively display the three-dimensional spatial distribution and dynamic changes of meteorological data, capture more complex weather patterns, and improve the accuracy and timeliness of weather forecasts. The three-dimensional training data also helps to simulate meteorological changes, that is, the changes in weather in three-dimensional space. Combined with actual topographic maps, it can more accurately predict meteorological changes in the target area. For example, in plain areas, it may bring relatively uniform wind and rain weather, while in mountainous areas, the local climate may be changeable due to the influence of terrain, such as strong winds, rainfall or haze.

[0022] In one implementation, terrain distribution maps and space weather data are input into the weather forecast model together. The terrain distribution map can accurately reflect the terrain and geomorphic features of the earth's surface, such as mountains, rivers, lakes, plains, etc. These features have an important impact on the direction and distribution of airflow. By combining terrain distribution maps with space weather data, the model can more accurately simulate and analyze meteorological phenomena such as atmospheric circulation and precipitation distribution, thereby improving the accuracy of weather forecasts.

[0023] In one implementation, weather classification can be divided into multiple types according to meteorological characteristics and changing patterns, including clear and cloudless weather, cloudy days with dense clouds, rainy days with heavy precipitation, snowy days in cold seasons, typhoons formed by tropical cyclones, and liquid precipitation (such as rain), solid precipitation (such as snow, hail) and mixed precipitation (such as sleet) according to the form of precipitation, which together constitute complex and changeable weather phenomena. According to weather classification, the specific weather type can be determined, and whether it is a meteorological disaster can be determined according to its impact range and intensity.

[0024] In one embodiment, step S101 includes: The satellite observation data is divided into a target sub-region set according to a preset interval, and the ground meteorological station data and the high-altitude meteorological observation data are spatially interpolated for any target sub-region according to a preset height to obtain a spatial distribution map; Arranging the spatial distribution graphs in time sequence to obtain a time series graph group; the time series graph group includes the spatial distribution graph at any time; The meteorological attribute information is added to the time series graph group to obtain training data; the meteorological attribute information includes: air temperature, air humidity, wind direction and wind speed.

[0025] In one implementation, ground meteorological station data include basic meteorological elements such as air temperature, air pressure, air humidity, wind direction and speed, precipitation, surface temperature, snow accumulation, clouds, visibility, and weather phenomena; high-altitude meteorological observation data include basic meteorological observation elements of specified layers such as air temperature, air pressure, wind speed and direction, potential height, and dew point temperature; satellite observation data include radiation brightness temperature, backscattering, cloud top temperature, ground temperature, average equivalent blackbody brightness temperature, ground incident solar radiation, water vapor content in the upper and middle troposphere, cloud cover, average snow cover, sandstorm monitoring and other meteorological information.

[0026] In one implementation, by spatially interpolating ground meteorological station data, high-altitude meteorological observation data, and satellite observation data according to the geographical location of the target area, a spatial distribution map of meteorological data in the target area can be generated. Spatial interpolation is to extrapolate meteorological data from sources such as ground meteorological stations, high-altitude meteorological observations, and satellites according to the vertical distribution based on the geographical location relationship to generate meteorological changes in the target area's meteorological elements that are continuously distributed over time. The above steps not only integrate meteorological data from multiple sources, but also ensure the comprehensiveness and accuracy of the data. These spatial distribution maps are arranged in chronological order into a time series map group, which provides a continuous and complete sequence of observation data for the training of weather models, thereby improving the accuracy of predictions of future weather conditions.

[0027] In one implementation, the satellite observation data is divided according to preset intervals to obtain a target sub-area set, that is, the target area is divided into multiple target sub-areas, each target sub-area corresponds to a collection point for ground meteorological station data and high-altitude meteorological observation data; on the basis of generating a spatial distribution map, meteorological attribute information is further added, such as: air temperature, air humidity, wind direction and wind speed, that is, the location of the data is first determined, and then the specific data is filled into the corresponding position, that is, the meteorological attribute information is added to the time series graph group; training data for training weather models is formed. The above steps enrich the dimension of the data set. Through the time series graph group, the model can capture the dynamic process of meteorological elements changing over time, thereby more effectively learning and simulating the complexity of the atmospheric system.

[0028] In one embodiment, before step S102, the following steps are included: Obtain historical training data and historical terrain distribution maps of the target area, and input the historical training data and historical terrain distribution maps into the machine learning model for training in time series; The output of the machine learning model is used as the target model parameter, and the parameters of the fully connected layer, convolutional layer, and recurrent layer of the machine learning model are updated; The machine learning model with updated parameters is used as a weather prediction model; the output of the weather prediction model is the training data of the target sub-area in the future time period.

[0029] In one implementation, the output of the weather forecast model is the training data of the target sub-region in the future time period, that is, the spatial meteorological data and meteorological attribute information in the future time period; the historical terrain distribution map is the terrain map of the target area, such as: hills or plains and other terrains; by inputting the historical training data (including ground meteorological station data, high-altitude meteorological observation data and satellite observation data, which also contain atmospheric circulation information) and the historical terrain distribution map into the machine learning model in time series, and comprehensively considering the impact of atmospheric circulation on the weather in the local area for comprehensive training, the model can deeply learn the relationship between terrain factors and meteorological changes. In this way, the effect of terrain characteristics on weather conditions can be more accurately reflected in the prediction process, so that the accuracy of weather forecasts is significantly improved, and weather changes in local areas can be better predicted.

[0030] In one implementation, during the training process, the output of the machine learning model is used as the target model parameters, and then the parameters of the fully connected layer, convolutional layer, and recurrent layer of the model are updated. This step not only optimizes the internal structure of the model, but also improves the model's processing ability and prediction performance for time series data. The machine learning model with updated parameters is used as a weather forecast model, which can more accurately predict meteorological changes in the future time period, including key meteorological elements such as temperature, humidity, and precipitation.

[0031] In one implementation, Dense Layers: Each neuron in these layers is connected to all neurons in the previous layer. Linear and nonlinear relationships in the data are learned and represented through weights and bias parameters. During training, the weights and bias parameters are updated according to the gradient of the loss function to minimize the prediction error. Convolutional Layers: Convolutional layers are very effective when processing images or grid data. Local features are extracted by sliding the convolution kernel (or filter) over the input data and calculating the dot product between the convolution kernel and the input data. In weather forecasting, convolutional layers can be used to extract the spatial distribution characteristics of meteorological elements. During training, the weights of the convolution kernels are updated. Recurrent Layers: Recurrent layers (such as LSTM or GRU) are very effective when processing sequence data, such as time series data. In weather forecasting, recurrent layers can be used to capture the changes in meteorological elements over time. During training, the weights and states of the recurrent layers are updated.

[0032] In one embodiment, step S102 further includes: Step 1: Detect the update status of the training data, detect the timestamp of the data update and determine the data update frequency, and calculate the data update coefficient; Step 2: Perform data integrity check on the training data, compare the weather classification with the historical data to determine whether there is data missing, and calculate the data missing coefficient; Step 3: Perform outlier detection on the training data, determine whether there is tampering, and calculate the data tampering coefficient; Step 4: Calculate the data evaluation coefficient based on the data update coefficient, data missing coefficient and data tampering coefficient, and evaluate the training data based on the data evaluation coefficient.

[0033] In one implementation, the data update coefficient is the frequency of data updates within a preset time period. The larger the data update coefficient, the more data updates are performed at this time. By detecting the update status of the training data, the system can accurately track the timestamp of the data update and determine the data update frequency accordingly. The calculated data update coefficient reflects the timeliness of the data. Ensuring the high timeliness of training data is crucial to improving the accuracy of weather forecast models, because the latest data can better reflect the current atmospheric state and changing trends, thereby helping the model make predictions that are closer to reality.

[0034] In one implementation, the data missing coefficient refers to whether there are gaps in the data and the proportion of the number of gaps to the total number. The larger the data missing coefficient, the more data gaps there are at this time. The data integrity detection link can timely discover possible gaps in the data set by comparing weather classification with historical data. The calculated data missing coefficient intuitively shows the completeness of the data set. A complete data set is the basis for model training. Data missingness will affect the prediction performance of the model, so it is necessary to ensure the comprehensiveness and consistency of the training data.

[0035] In one implementation, the data tampering coefficient is the ratio of the number of tampered data to the total number. The larger the data tampering coefficient, the more the data has been tampered with. Outlier detection and tampering judgment are key steps to ensure the quality of training data. By detecting abnormal fluctuations or unreasonable values ​​in the data and evaluating whether there is human tampering, the data tampering coefficient can be calculated to evaluate the authenticity and credibility of the data. High-credibility training data is a prerequisite for the model to make accurate predictions, because false or misleading data may mislead the training direction of the model.

[0036] In one embodiment, step 4 includes: Calculation formula of coefficient through data evaluation Calculate the data evaluation coefficient, and evaluate the training data according to the data evaluation coefficient; Among them, F is the data evaluation score, G is the data update coefficient, C is the data tampering coefficient, Q is the data missing coefficient, α and β are constant proportional coefficients, and both α and β are not zero.

[0037] In one implementation, the data evaluation coefficient is used to evaluate the integrity and timeliness of the training data. A higher data evaluation score indicates that the training data is more complete and more timely, that is, the data is up to date. When the data missing coefficient and the data tampering coefficient increase while the other coefficients remain unchanged, the training data will be missing and contain more data errors, so the data evaluation coefficient score will decrease. When the data update coefficient increases while the other coefficients remain unchanged, the data evaluation coefficient score will increase, indicating that the data is updated promptly and is timely.

[0038] In one embodiment, evaluating the training data according to the data evaluation coefficient includes: If the data evaluation coefficient is less than the data threshold, an alarm is issued and space weather data is collected again; If the data evaluation coefficient is not less than the data threshold, the training data is input into the weather prediction model for prediction.

[0039] In one implementation, when the data evaluation coefficient is less than the data threshold, triggering an alarm and re-updating the data can ensure the accuracy and reliability of the data and avoid prediction biases caused by data quality issues. Re-collecting data can correct errors or biases and improve data accuracy, thereby enhancing the performance of machine learning, enabling the model to better learn the patterns in the data and improve prediction accuracy and generalization capabilities.

[0040] In one implementation, if the data evaluation coefficient is not less than the data threshold, the weather classification data is directly input into the prediction model. The above steps can utilize high-quality data resources, improve the accuracy and timeliness of weather forecasts, and improve overall operational efficiency and effectiveness.

[0041] In one embodiment, step S103 includes: The wind direction in each weather forecast result is fitted to obtain the wind direction curve of the target area, and the cloud displacement in the satellite observation data is calculated based on the wind speed; Determine the cloud accumulation area in the target area based on the cloud displacement in the satellite observation data, and determine the weather type of the cloud accumulation area based on the ground meteorological station data and high-altitude meteorological observation data in the weather forecast results; The type of meteorological disaster is determined based on weather classification. If the impact range of the meteorological disaster type is greater than the area threshold and the intensity of the meteorological disaster type is greater than the intensity threshold, a meteorological disaster warning is issued based on the meteorological disaster type.

[0042] In one implementation, because the input to the model is the meteorological data of each target sub-area, the comprehensive situation of the entire target area can be obtained by combining and analyzing the data of each target sub-area; the wind direction curve of the target area is obtained by fitting the wind direction of each target sub-area, and this wind direction curve represents the direction of the wind in the future time period. Combined with the wind speed, the movement distance of the cloud layer, that is, the cloud layer displacement, can be calculated. The cloud layer can be obtained through satellite observation data; when the cloud layer in the target area accumulates, the specific weather type is determined according to the ground meteorological station data and the high-altitude meteorological observation data in the weather forecast results. For example: if the ground meteorological station data and the high-altitude meteorological observation data show that it is raining in the target area, and the rainfall per hour reaches or exceeds 16 mm, or the rainfall for 12 consecutive hours reaches or exceeds 30 mm, or the rainfall for 24 hours reaches or exceeds 50 mm, then the weather type of the target area is judged to be heavy rain.

[0043] In one implementation, the weather conditions in the target area are first accurately located based on the weather classification. When a certain weather phenomenon is detected to have a wide impact and abnormally high intensity, the corresponding meteorological disaster warning level is determined based on the established standards. For example, when the weather classification shows rainfall, and the rainfall intensity and range reach or exceed the preset threshold, it can be determined as a rainstorm warning, and the warning level can be further subdivided according to the actual situation, which improves the accuracy and timeliness of meteorological warnings.

[0044] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for predicting meteorological disasters based on weather typing, characterized in that: The method comprises: Acquire space weather data of the target area, and perform three-dimensional processing on the space weather data to obtain training data; the space weather data is data collected at different altitudes in the target area; Obtaining a terrain distribution map of a target area, and inputting the training data and the terrain distribution map into a weather prediction model for training to obtain a weather prediction result; The weather type of the target area is determined according to the weather forecast result, and meteorological disaster prediction is performed according to the weather type.

2. A method for predicting meteorological disasters based on weather typing according to claim 1, characterized in that: The space weather data includes: ground weather station data, high-altitude weather observation data and satellite observation data. The space weather data is processed three-dimensionally to obtain training data including: The satellite observation data is divided into a target sub-region set according to a preset interval, and the ground meteorological station data and the high-altitude meteorological observation data are spatially interpolated for any target sub-region according to a preset height to obtain a spatial distribution map; Arranging the spatial distribution graphs in time sequence to obtain a time series graph group; the time series graph group includes the spatial distribution graph at any time; Meteorological attribute information is added to the time series graph group to obtain training data; the meteorological attribute information includes: air temperature, air humidity, wind direction and wind speed.

3. The method for predicting meteorological disasters based on weather typing according to claim 1, characterized in that: Before inputting the training data and the terrain distribution map into the weather prediction model for training to obtain the weather prediction result, the method includes: Acquire historical training data and historical terrain distribution maps of the target area, and input the historical training data and the historical terrain distribution maps into the machine learning model for training in a time series; The output of the machine learning model is used as a target model parameter, and the parameters of the fully connected layer, the convolutional layer, and the recurrent layer of the machine learning model are updated; The machine learning model with updated parameters is used as a weather prediction model; the output of the weather prediction model is the training data of the target sub-area in the future time period.

4. The method for predicting meteorological disasters based on weather typing according to claim 1, characterized in that: Before inputting the training data into the weather prediction model for training to obtain the weather prediction result, the following steps are included: Performing update status detection on the training data, detecting the timestamp of data update and determining the data update frequency, and calculating the data update coefficient; Performing data integrity check on the training data, comparing the weather classification with historical data to determine whether there is data missing, and calculating the data missing coefficient; Performing outlier detection on the training data, determining whether tampering has occurred, and calculating a data tampering coefficient; A data evaluation coefficient is calculated according to the data update coefficient, the data missing coefficient and the data tampering coefficient, and the training data is evaluated according to the data evaluation coefficient.

5. A method for predicting meteorological disasters based on weather typing according to claim 4, characterized in that: The data evaluation coefficients calculated according to the data update coefficient, data missing coefficient and data tampering coefficient include: Calculation formula of coefficient through data evaluation The data evaluation coefficient is calculated; Among them, F is the data evaluation score, G is the data update coefficient, C is the data tampering coefficient, Q is the data missing coefficient, α and β are constant proportional coefficients, and both α and β are not zero.

6. A method for predicting meteorological disasters based on weather typing according to claim 4, characterized in that: Evaluating the training data according to the data evaluation coefficient includes: If the data evaluation coefficient is less than the data threshold, an alarm is issued and space weather data is collected again; If the data evaluation coefficient is not less than the data threshold, the training data is input into the weather prediction model for prediction.

7. The method for predicting meteorological disasters based on weather typing according to claim 1, characterized in that: Meteorological disaster prediction based on the weather classification includes: Fitting the wind directions in each weather forecast result to obtain a wind direction curve of the target area, and calculating the cloud displacement in the satellite observation data according to the wind speed; Determine the cloud accumulation area of ​​the target area according to the cloud displacement in the satellite observation data, and determine the weather type of the cloud accumulation area according to the ground meteorological station data and the high-altitude meteorological observation data in the weather forecast result; The type of meteorological disaster is determined according to the weather classification. If the impact range of the meteorological disaster type is greater than the area threshold and the intensity of the meteorological disaster type is greater than the intensity threshold, a meteorological disaster warning is issued according to the meteorological disaster type.

Images