A method for processing meteorological data

Abstract

The present invention relates to the field of meteorological data processing technology, and in particular to a meteorological data processing method. The method comprises the following steps: collecting meteorological data, such as temperature, humidity, etc., from multiple data sources, and preprocessing using filtering technology; extracting key meteorological features through data fusion algorithms and data mining; mapping meteorological data in three-dimensional space using geographic information system technology to construct a three-dimensional meteorological data map; using a graph neural network algorithm to construct a meteorological prediction model, identify meteorological conditions and trends, and conduct a comprehensive evaluation in combination with an expert system to provide early warning and decision support for relevant personnel; providing a visual interactive interface to present meteorological data in the form of charts and reports, and providing a feedback mechanism. By efficiently integrating and analyzing meteorological data, the present invention can not only provide accurate meteorological forecasts and effective decision support, but also improve the availability of meteorological data and user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological data processing, and in particular to a meteorological data processing method. Background Art

[0002] With the rapid development of information technology, digitalization has become a prominent feature of modern society. As an important place to popularize scientific knowledge and improve the public's scientific literacy, digital science museums play a unique and important role. By utilizing advanced information technology, digital science museums can not only display complex meteorological data, but also allow visitors to intuitively understand meteorological knowledge and the impact of climate change through interactive exhibitions. However, existing meteorological data processing often cannot fully meet the requirements of digital science museums. At present, there are still the following problems: existing methods fail to fully utilize the fusion of multi-source data, resulting in the inability to effectively integrate information from different data sources and it is difficult to accurately extract key meteorological features; existing technologies are not accurate enough in meteorological forecasting and cannot efficiently identify meteorological conditions and trends. Summary of the invention

[0003] In order to solve the above problems, the present invention provides a meteorological data processing method.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] A method for processing meteorological data comprises the following steps:

[0006] Collect meteorological data in real time through multiple data sources, and pre-process the meteorological data using filtering technology; the meteorological data includes temperature, humidity, wind speed, air pressure, precipitation and satellite cloud images;

[0007] The multi-source meteorological data are fused by a data fusion algorithm, and key meteorological features are extracted by data mining; the key meteorological features include temperature trends, humidity patterns, and air pressure changes;

[0008] Based on geographic information system technology, the fused meteorological data is mapped into three-dimensional space to construct a three-dimensional meteorological data map;

[0009] Building a weather forecast model based on a graph neural network algorithm to identify weather conditions and trends; the weather forecast model is used to provide weather forecasts by analyzing complex relationships and patterns between weather data;

[0010] Combined with the expert system, the analysis results are comprehensively evaluated and optimized, and meteorological warnings and decision-making support are provided to relevant personnel;

[0011] Provides a visual interactive interface, converts meteorological data into charts and reports, and provides a feedback mechanism.

[0012] Furthermore, the data sources include ground weather stations, weather satellites, weather radars and sensors carried by drones; and the filtering techniques include Kalman filtering and / or moving average filtering.

[0013] Furthermore, the data fusion algorithm specifically adopts a multi-level data fusion framework to fuse multi-source meteorological data, and the multi-level fusion framework includes a primary fusion layer, an intermediate fusion layer, and an advanced fusion layer;

[0014] The primary fusion layer merges homogeneous data by arithmetic averaging or weighted averaging; the intermediate fusion layer integrates data of different measurement units or magnitudes by principal component analysis or Kalman filtering; the advanced fusion layer uses machine learning algorithms to integrate the relationships between various types of data and output the final fused data set.

[0015] Furthermore, based on the geographic information system technology, the fused meteorological data is mapped into a three-dimensional space to construct a three-dimensional meteorological data map, including the following steps:

[0016] Use geographic information system technology to spatially locate the received fused meteorological data;

[0017] By using 3D modeling technology, the located meteorological data is converted into visualization elements in 3D space, and the meteorological data is presented in different colors, shapes or sizes in the 3D map;

[0018] Applying 3D interpolation and smoothing techniques, a continuous 3D meteorological field is generated based on known meteorological data points, providing reasonable meteorological information forecasts in data-sparse areas;

[0019] Combined with dynamic simulation technology, the time series dynamic display of meteorological data can be realized;

[0020] Integrated interactive functions allow users to view three-dimensional weather maps from multiple angles and scales through zooming, rotating and panning, and also support clicking on specific meteorological elements to query detailed data.

[0021] Furthermore, the weather forecasting model is constructed according to the graph neural network algorithm to identify weather conditions and trends, including the following steps:

[0022] Receive fused meteorological data;

[0023] Using the structure of graph neural network, the geographical location and time series information in the meteorological data are encoded into nodes and edges, and a graph representation of the meteorological data is constructed, where nodes represent specific meteorological observation points and edges represent the spatial or temporal relationship between observation points.

[0024] Through the forward propagation of the graph neural network, feature learning is performed on the meteorological data graph to capture the complex spatial and temporal dependencies between meteorological data;

[0025] Based on the learned meteorological features, a meteorological prediction model is constructed using a graph neural network. The meteorological prediction model predicts and analyzes meteorological conditions based on the spatial relationship and temporal evolution of meteorological data.

[0026] Combine expert knowledge and historical data to continuously train and optimize weather forecast models.

[0027] Furthermore, the formula used in the weather forecast model is as follows:

[0028]

[0029] Among them, Y represents the forecast output of meteorological conditions, including the forecast values ​​of key meteorological characteristics such as future temperature trends, humidity patterns, and air pressure changes; X i represents the feature vector extracted from the fused meteorological data; W i represents the model parameters, i.e., weights, which are used to adjust the influence of the corresponding meteorological data Xx in the prediction model; b represents a constant, which is used to adjust the model output; N represents the number of meteorological data considered in the fused meteorological data; σ represents the activation function, which is used to increase the nonlinear fitting ability of the model.

[0030] Furthermore, the three-dimensional modeling technology specifically includes using high-precision terrain data to generate a three-dimensional grid model of the terrain, applying physical rendering technology to simulate the three-dimensional distribution of meteorological data, and using volume rendering technology to stereoscopically visualize cloud maps; and through a real-time dynamic update mechanism, updating the three-dimensional meteorological map according to the latest meteorological data.

[0031] Furthermore, the expert system is specifically based on expert knowledge and experience in the field of meteorology, and is used to interpret and verify the analysis results of meteorological data. At the same time, according to the feedback from the expert system and the output results of the meteorological prediction model, the parameters of the meteorological prediction model are dynamically adjusted, and meteorological warnings and decision support are generated for users through the interpretation and reasoning provided by the expert system.

[0032] Furthermore, the visualization interaction interface is specifically based on an interactive 3D visualization tool of WebGL, which uses virtual reality technology to display meteorological data and realizes three-dimensional display of data through user interaction.

[0033] The beneficial effects of the present invention are:

[0034] The present invention can ensure the data quality of the input model by collecting meteorological data from multiple data sources in real time and preprocessing it using filtering technology, thereby improving the accuracy of meteorological forecasts. The data fusion algorithm further integrates these multi-source data and provides more comprehensive and accurate meteorological information through comprehensive analysis. The use of data mining technology to extract key meteorological features, such as temperature trends, humidity patterns, and air pressure changes, makes the analysis of meteorological data more in-depth and helps to understand complex meteorological phenomena.

[0035] Based on geographic information system technology, the present invention maps the fused meteorological data into three-dimensional space and constructs a three-dimensional meteorological data map, which not only enhances the intuitiveness of the data, but also improves the understanding and analysis capabilities of the spatial distribution characteristics of meteorological data. The meteorological forecasting model constructed using the graph neural network algorithm can accurately identify meteorological conditions and trends, and provide scientific predictions for future meteorological changes by analyzing the complex relationships and patterns between meteorological data. Combined with the expert system for comprehensive evaluation and optimization, it can provide a scientific basis for meteorological decision-making, provide meteorological warnings for relevant personnel, enhance the ability to respond to extreme meteorological events, and reduce possible losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of a meteorological data processing method of the present invention.

[0037] Figure 2 It is a flowchart of step S3 provided in an embodiment of the present invention.

[0038] Figure 3 It is a flowchart of step S4 provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] See also Figure 1-3 As shown, the present invention relates to a meteorological data processing method.

[0040] Example

[0041] A method for processing meteorological data comprises the following steps:

[0042] S1: Meteorological data are collected in real time through multiple data sources, and filtering technology is used to pre-process the meteorological data; the meteorological data include temperature, humidity, wind speed, air pressure, precipitation and satellite cloud images; the data sources include ground meteorological stations, meteorological satellites, meteorological radars and sensors carried by drones; the filtering technology includes Kalman filtering and / or moving average filtering.

[0043] Specifically, a variety of sensors, such as thermometers and hygrometers, anemometers, barometers, and rain gauges, are deployed at ground-based meteorological stations located in different geographical locations to monitor and record meteorological data in real time. At the same time, meteorological satellites are used to obtain large-scale meteorological conditions, including cloud maps and atmospheric temperature distribution. Weather radars are used to capture dynamic images of precipitation and storms, while drones fly at low altitudes in specific areas to collect fine-grained meteorological data.

[0044] The Kalman filter algorithm is applied to the data output by each sensor for dynamic estimation and correction; the accuracy of the data is improved by removing the noise introduced by factors such as equipment errors and environmental interference. For time-varying data such as temperature and humidity, moving average filtering is applied to smooth short-term fluctuations and retain long-term trends.

[0045] S2: fusing multi-source meteorological data through a data fusion algorithm and extracting key meteorological features through data mining; the key meteorological features include temperature trends, humidity patterns, and air pressure changes;

[0046] The data fusion algorithm specifically adopts a multi-level data fusion framework to fuse multi-source meteorological data. The multi-level fusion framework includes a primary fusion layer, an intermediate fusion layer and an advanced fusion layer.

[0047] The primary fusion layer merges homogeneous data by arithmetic averaging or weighted averaging; the intermediate fusion layer integrates data of different measurement units or magnitudes by principal component analysis or Kalman filtering; the advanced fusion layer uses machine learning algorithms to integrate the relationships between various types of data and output the final fused data set.

[0048] Specifically, the primary fusion layer is used to merge homogeneous data. The arithmetic mean is calculated for the data collected by different sensors such as temperature and humidity at the same location to reduce random errors. The weighted average method is used to assign different weights to data from different sources based on data quality. For example, the weight of weather station data may be higher than that of data collected by drones because its stability and accuracy are usually higher. The intermediate fusion layer is used to integrate data with different units or magnitudes. Principal component analysis is used for data with different dimensions (such as temperature and air pressure). PCA is used to extract the main variables and patterns in the data while reducing the dimensions for further analysis. Kalman filtering is used for the fusion of time series data, especially in dynamic environments. The quality of the integrated data is improved by predicting and updating data with different units or magnitudes. The advanced fusion layer analyzes the interaction between different data through machine learning algorithms (such as random forests or deep learning), comprehensively considers all available information, and outputs the final fused data set. Using data mining techniques, key features such as temperature trends, humidity patterns, and air pressure changes are extracted from the fused data. These features are essential for understanding climate conditions and predicting future trends.

[0049] S3: Based on geographic information system technology, the fused meteorological data is mapped into three-dimensional space to construct a three-dimensional meteorological data map;

[0050] Wherein, step S3 specifically includes the following steps:

[0051] S31: Use geographic information system technology to spatially locate the received fused meteorological data;

[0052] S32: The located meteorological data is converted into visualization elements in three-dimensional space through three-dimensional modeling technology, and the meteorological data is displayed in different colors, shapes or sizes in the three-dimensional map; the three-dimensional modeling technology specifically includes generating a three-dimensional grid model of the terrain using high-precision terrain data, simulating the three-dimensional distribution of meteorological data using physical rendering technology, and stereoscopically visualizing cloud images using volume rendering technology; and updating the three-dimensional meteorological map according to the latest meteorological data through a real-time dynamic update mechanism;

[0053] Specifically, import meteorological data into the 3D modeling software. This step may require converting the data format or using plug-ins to support data reading. Obtain high-precision terrain data (DEM) that matches the project's geographic area from public data sources. Import the DEM data into the 3D modeling software and use software tools to convert it into a 3D terrain grid model. This process may involve adjusting and optimizing terrain details to meet different visualization needs. According to the needs of meteorological data display, make necessary adjustments to the generated 3D terrain model, such as adjusting the perspective, scaling, or cutting the terrain to better display the meteorological data. According to different types of meteorological data (such as temperature, humidity, precipitation), define appropriate visualization parameters such as color gradient, transparency, size, etc. Use the physical rendering engine of the 3D modeling software to set the material and map of the meteorological data according to the defined parameters to realistically simulate the visual effects of meteorological phenomena. For example, volumetric cloud rendering technology can be used to simulate the distribution and density of clouds. For meteorological data that needs to show the changing process, such as moving clouds or rainfall, use particle systems or dynamic simulation technology to create dynamic effects. Set up a background service to listen to meteorological data updates, and automatically trigger the update process of the 3D model once new data is detected. This may include re-importing data, re-rendering of models, etc. Define different update strategies according to the update frequency and importance of meteorological data. For example, for meteorological phenomena that change in real time (such as heavy rain movement), you may need to set a shorter update cycle, while for data that changes more slowly (such as seasonal temperature changes), the update cycle can be relatively long.

[0054] S33: Apply three-dimensional interpolation and smoothing techniques to generate a continuous three-dimensional meteorological field based on known meteorological data points, providing reasonable meteorological information forecasts in data-sparse areas;

[0055] It should be noted that for known meteorological data points, three-dimensional interpolation techniques (such as Kriging interpolation and radial basis function interpolation) are used to generate a continuous three-dimensional meteorological field. This helps to provide continuous and reasonable meteorological information forecasts in areas with sparse data. Smoothing techniques (such as Gaussian smoothing) are used to optimize the interpolation results, reduce data noise, and improve the visual effect of meteorological information.

[0056] S34: Combined with dynamic simulation technology, the time series dynamic display of meteorological data is realized;

[0057] S35: Integrated interactive functions allow users to view three-dimensional weather maps from multiple angles and scales through zooming, rotating and panning operations, and also support clicking on specific meteorological elements to query detailed data.

[0058] S4: constructing a meteorological prediction model based on a graph neural network algorithm to identify meteorological conditions and trends; the meteorological prediction model is used to provide meteorological predictions by analyzing complex relationships and patterns between meteorological data;

[0059] Wherein, step S4 specifically includes the following steps:

[0060] S41: receiving fused meteorological data;

[0061] S42: Using the structure of a graph neural network, the geographical location and time series information in the meteorological data are encoded into nodes and edges, and a graph representation of the meteorological data is constructed, where the nodes represent specific meteorological observation points and the edges represent the spatial or temporal relationship between the observation points;

[0062] It should be noted that each meteorological station not only contains basic geographic location (such as longitude, latitude) and time series information, but also can encode more advanced features, such as altitude, local terrain type (mountainous area, plain, etc.), and even statistical information of the historical meteorological data of the station (such as average temperature, humidity range, etc.). Such advanced encoding provides rich contextual information for the model and helps capture complex patterns in meteorological data. For time information, in addition to directly using timestamps, periodic encoding (such as sine and cosine encoding) can be used to represent the time of day and the date of the year to capture the seasonality and daily variation of meteorological conditions. This method helps the model understand the role of time in meteorological changes. When defining edges, weights can be dynamically adjusted according to the actual geographical distance and meteorological similarity between meteorological stations. For example, stations with similar meteorological conditions (even if they are geographically far away) can be given higher weights, reflecting the characteristics of the spatial propagation of meteorological phenomena.

[0063] S43: Through the forward propagation of the graph neural network, feature learning is performed on the meteorological data graph to capture the complex spatial and temporal dependencies between meteorological data;

[0064] S44: constructing a meteorological prediction model using a graph neural network according to the learned meteorological features, wherein the meteorological prediction model predicts and analyzes meteorological conditions based on the spatial relationship and temporal evolution of meteorological data;

[0065] S45: Continuously train and optimize weather forecast models by combining expert knowledge and historical data;

[0066] Furthermore, the formula used in the weather forecast model is as follows:

[0067]

[0068] Among them, Y represents the forecast output of meteorological conditions, including the forecast values ​​of key meteorological characteristics such as future temperature trends, humidity patterns, and air pressure changes; X i represents the feature vector extracted from the fused meteorological data; W i Represents the model parameters, i.e., weights, used to adjust the corresponding meteorological data X i The influence of in the prediction model; b represents a constant, which is used to adjust the model output; N represents the number of meteorological data considered in the fused meteorological data; σ represents an activation function, which is used to increase the nonlinear fitting ability of the model.

[0069] S5: Combined with the expert system, comprehensive evaluation and optimization of the analysis results are carried out, and meteorological warning and decision-making support are provided to relevant personnel;

[0070] Among them, the expert system is specifically based on expert knowledge and experience in the field of meteorology, and is used to interpret and verify the analysis results of meteorological data. At the same time, according to the feedback of the expert system and the output results of the meteorological prediction model, the parameters of the meteorological prediction model are dynamically adjusted. Through the explanation and reasoning provided by the expert system, meteorological warnings and decision support are generated for users.

[0071] Specifically, the knowledge and experience of experts in the field of meteorology are collected and organized, including rules, facts, assumptions and prediction models. This knowledge can be about meteorological pattern recognition, climate change trend analysis, warning standards for extreme weather events, etc. Reasoning algorithms that simulate the expert thinking process include forward reasoning (from known facts to conclusions) and reverse reasoning (deriving required conditions from expected results). These reasoning mechanisms can help the system make reasonable judgments and predictions based on the current meteorological data analysis results. The output of the meteorological prediction model is evaluated by the expert system, and the prediction results are compared with historical data and known meteorological events to verify the accuracy and reliability of the prediction. According to the evaluation results and feedback of the expert system, the parameters of the meteorological prediction model are adjusted, such as improving the model algorithm, adjusting the data weight, etc., to improve the prediction performance of the model. Based on the model prediction and the comprehensive evaluation of the expert system, meteorological warning information is generated and updated in real time, and warnings are issued to relevant departments and the public for impending extreme weather events (such as typhoons, heavy rains, and high temperature waves). Based on the interpretation and reasoning of the expert system, data-based suggestions and possible response measures are provided to decision makers to help them formulate emergency plans and resource allocation strategies.

[0072] S6: Provide a visual interactive interface to convert meteorological data into charts and reports and provide a feedback mechanism; the visual interactive interface is specifically based on an interactive 3D visualization tool of WebGL, which uses virtual reality technology to display meteorological data and realizes three-dimensional display at the data level through user interaction.

[0073] Specifically, a browser-based interactive visualization platform is developed using WebGL technology, which can display complex three-dimensional meteorological data maps and models. Users can view these visualizations in any browser that supports WebGL. WebGL technology can use hardware acceleration to provide high-performance three-dimensional graphics rendering, allowing users to explore and analyze large-scale meteorological data sets in real time. Through virtual reality (VR) technology, a more immersive meteorological data visualization experience is created. After wearing a VR helmet, users can enter a three-dimensional meteorological data environment and intuitively feel and analyze meteorological phenomena. VR technology provides a new way of data interaction and visualization, allowing users to gain a deeper understanding of meteorological data from a new perspective, especially for complex climate models and forecasting analysis. Charts and reports are dynamically generated based on user query conditions and data selections. These charts can be the results of time series analysis, visualization of climate trends, or analysis of specific meteorological events. Through dynamic generation, it can be ensured that users always get the latest data analysis results. In addition, users can customize the content of charts and reports according to their needs, which enhances the personalized experience. A user feedback mechanism is integrated into the visualization platform, and users can ask queries, share observations, or provide feedback through the platform. These interactive features can help improve the presentation of data and enhance user experience.

[0074] In summary, the present invention can ensure the data quality of the input model by collecting meteorological data from multiple data sources in real time and preprocessing it using filtering technology, thereby improving the accuracy of meteorological forecasts. A multi-level data fusion framework is used to integrate meteorological information from multiple data sources such as ground meteorological stations, meteorological satellites, meteorological radars, and drones, which can provide a more comprehensive and integrated view of meteorological data. Extracting key features (such as temperature trends, humidity patterns, and air pressure changes) from the fused meteorological data through data mining technology helps to deeply understand climate conditions and predict future trends.

[0075] The present invention combines geographic information system technology and 3D modeling technology to map meteorological data into 3D space, and provides an intuitive and dynamic meteorological data visualization experience through high-precision visual presentation methods (such as physical rendering and volume rendering technology). The meteorological forecast model constructed using the graph neural network algorithm can identify the complex relationships and patterns between meteorological data, provide accurate and timely meteorological forecasts, and thus provide a strong scientific basis for meteorological warning and decision support.

[0076] The present invention provides users with a platform for intuitive operation, multi-angle viewing and in-depth analysis of meteorological data by developing an interactive 3D visualization tool based on WebGL and virtual reality technology, greatly enhancing the user experience. Combined with the output of the expert system and the meteorological forecast model, it can provide relevant personnel with meteorological warning information and decision support based on data and professional knowledge, helping them better cope with extreme weather events and climate change.

[0077] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A meteorological data processing method, characterized in that: The following steps are involved: Collect meteorological data in real time through multiple data sources, and pre-process the meteorological data using filtering technology; the meteorological data includes temperature, humidity, wind speed, air pressure, precipitation and satellite cloud images; The multi-source meteorological data are fused through data fusion algorithm, and key meteorological features are extracted through data mining; Said key meteorological characteristics include temperature trends, humidity patterns and changes in air pressure; Based on geographic information system technology, the fused meteorological data is mapped into three-dimensional space to construct a three-dimensional meteorological data map; Building a weather forecast model based on a graph neural network algorithm to identify weather conditions and trends; the weather forecast model is used to provide weather forecasts by analyzing complex relationships and patterns between weather data; Combined with the expert system, the analysis results are comprehensively evaluated and optimized, and meteorological warnings and decision-making support are provided to relevant personnel; Provide a visual interactive interface to convert meteorological data into charts and reports, and provide a feedback mechanism; The data fusion algorithm specifically adopts a multi-level data fusion framework to fuse multi-source meteorological data. The multi-level data fusion framework includes a primary fusion layer, an intermediate fusion layer and an advanced fusion layer. The primary fusion layer combines homogeneous data by arithmetic averaging or weighted averaging; the intermediate fusion layer integrates data of different units or magnitudes by principal component analysis or Kalman filtering; the advanced fusion layer integrates the relationships between various types of data using machine learning algorithms and outputs the final fused data set; Based on the geographic information system technology, the fused meteorological data is mapped into a three-dimensional space to construct a three-dimensional meteorological data map, including the following steps: Use geographic information system technology to spatially locate the received fused meteorological data; By using 3D modeling technology, the located meteorological data is converted into visualization elements in 3D space, and the meteorological data is presented in different colors, shapes or sizes in the 3D map; Applying 3D interpolation and smoothing techniques, a continuous 3D meteorological field is generated based on known meteorological data points, providing reasonable meteorological information forecasts in data-sparse areas; Combined with dynamic simulation technology, the time series dynamic display of meteorological data can be realized; Integrated interactive functions allow users to view 3D weather maps from multiple angles and scales through zooming, rotating and panning operations, and also support clicking on specific weather elements to query detailed data; The method of building a weather forecast model based on a graph neural network algorithm to identify weather conditions and trends includes the following steps: Receive fused meteorological data; Using the structure of graph neural network, the geographical location and time series information in the meteorological data are encoded into nodes and edges, and a graph representation of the meteorological data is constructed, where nodes represent specific meteorological observation points and edges represent the spatial or temporal relationship between observation points. Through the forward propagation of the graph neural network, feature learning is performed on the meteorological data graph to capture the complex spatial and temporal dependencies between meteorological data; Based on the learned meteorological features, a meteorological prediction model is constructed using a graph neural network. The meteorological prediction model predicts and analyzes meteorological conditions based on the spatial relationship and temporal evolution of meteorological data. Continuously train and optimize weather forecast models by combining expert knowledge and historical data; The formula used in the weather forecast model is as follows: ； Among them, Y represents the forecast output of meteorological conditions, including the forecast values ​​of key meteorological characteristics of future temperature trends, humidity patterns and air pressure changes; represents the feature vector extracted from the fused meteorological data; Represents the model parameters, i.e. weights, used to adjust the corresponding meteorological data The influence in the prediction model; b represents a constant used to adjust the model output; N represents the number of meteorological data considered in the fused meteorological data; Represents the activation function, which is used to increase the nonlinear fitting ability of the model; The three-dimensional modeling technology specifically includes using high-precision terrain data to generate a three-dimensional grid model of the terrain, applying physical rendering technology to simulate the three-dimensional distribution of meteorological data, and using volume rendering technology to stereoscopically visualize cloud maps; and through a real-time dynamic update mechanism, updating the three-dimensional meteorological map according to the latest meteorological data.

2. A meteorological data processing method according to claim 1, characterized in that: The data sources include ground weather stations, weather satellites, weather radars and sensors carried by unmanned aerial vehicles; the filtering technology includes Kalman filtering and / or moving average filtering.

3. A meteorological data processing method according to claim 1, characterized in that: The expert system is specifically based on expert knowledge and experience in the field of meteorology, and is used to interpret and verify the analysis results of meteorological data. At the same time, according to the feedback of the expert system and the output results of the meteorological prediction model, the parameters of the meteorological prediction model are dynamically adjusted. Through the interpretation and reasoning provided by the expert system, meteorological warnings and decision support are generated for users.

4. A meteorological data processing method according to claim 1, characterized in that: The visualization interaction interface is specifically based on an interactive 3D visualization tool of WebGL, which uses virtual reality technology to display meteorological data and realizes three-dimensional display at the data level through user interaction.

Images