A weather forecast optimization method and system based on complex terrain
By collecting and analyzing topographic and meteorological data in areas with complex terrain, a multi-level, multi-resolution weather forecasting model is constructed, which solves the problem of neglecting the influence of terrain in traditional weather forecasting methods and achieves more accurate weather forecasts.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA SOUTHERN POWER GRID COMPANY
- Filing Date
- 2023-12-18
- Publication Date
- 2026-06-23
Smart Images
Figure CN117973590B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of weather forecasting, and specifically relates to a weather forecasting optimization method and system based on complex terrain. Background Art
[0002] Complex terrain has a significant impact on the accuracy of weather forecasting. Geographical factors such as terrain undulation, mountain range orientation, and plateau distribution will all affect meteorological factors such as air flow, temperature, and humidity, making the weather forecasting method based on traditional numerical models perform poorly in complex terrain areas. In complex terrain, obtaining and analyzing meteorological data faces huge challenges. For example, effects such as mountain blockage and air flow refraction cause errors in the collection, processing, and analysis of meteorological data. In addition, due to the complex terrain, conventional meteorological models may not be able to accurately capture the weather changes in these areas.
[0003] In the process of implementing the inventive technical solution in the embodiments of this application, it is found that the above technology has at least the following technical problems:
[0004] Traditional weather forecasting methods ignore the specific impact of terrain on weather, especially the problem of large errors in complex terrain areas. Summary of the Invention
[0005] This application mainly solves the problem that traditional weather forecasting methods ignore the specific impact of terrain on weather, especially the problem of large errors in complex terrain areas.
[0006] In view of the above problems, this application provides a weather forecasting optimization method and system based on complex terrain. In the first aspect, this application provides a weather forecasting optimization method based on complex terrain, and the method includes: using meteorological data collection equipment to collect terrain data within a target area and historical meteorological data corresponding to the terrain data; establishing a terrain impact assessment model; inputting the terrain data and the historical meteorological data into the terrain impact assessment model to obtain N terrain impact factors, where the N terrain factors include M marker information, and M < N; using the N terrain impact factors as input data respectively, and combining real-time meteorological data to construct a multi-level and multi-resolution weather forecasting model, the multi-level and multi-resolution weather forecasting model includes multiple sub-models, and each of the sub-models corresponds to one of the N terrain impact factors; generating weather forecasting information through the weather forecasting model.
[0007] Second aspect, the present application provides a weather forecast optimization system based on complex terrain. The system includes: a historical data acquisition module for collecting terrain data and historical weather data corresponding to the terrain data within a target area by using meteorological data collection devices; an evaluation model establishment module for establishing a terrain impact evaluation model; a terrain impact factor acquisition module for inputting the terrain data and the historical weather data into the terrain impact evaluation model to obtain N terrain impact factors, where the N terrain factors include M marking information, and M < N; a weather forecast model construction module for using the N terrain impact factors as input data respectively and combining real-time meteorological data to construct a multi-level and multi-resolution weather forecast model, where the multi-level and multi-resolution weather forecast model includes multiple sub-models, and each of the sub-models corresponds to one of the N terrain impact factors; a weather forecast information generation module for generating weather forecast information through the weather forecast model.
[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages:
[0009] The present application provides a weather forecast optimization method and system based on complex terrain, relating to the technical field of weather forecast. The method includes: collecting historical weather data by using meteorological data collection devices, establishing a terrain impact evaluation model, then obtaining N terrain impact factors, combining real-time meteorological data to construct a multi-level and multi-resolution weather forecast model, and then generating weather forecast information through the weather forecast model.
[0010] The present application mainly solves the problem that traditional weather forecast methods ignore the specific impact of terrain on weather, especially large errors will occur in complex terrain areas. By analyzing a large amount of geographical information data and combining it with the weather forecast model, the technical effect of improving the accuracy of the forecast is achieved.
[0011] The above description is only an overview of the technical solutions of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the specific embodiments of the present application are hereinafter specifically exemplified. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0013] Figure 1 This application provides a schematic flowchart of a weather forecast optimization method based on complex terrain.
[0014] Figure 2 This application provides a schematic flowchart of a method for constructing a terrain impact assessment model in a weather forecast optimization method based on complex terrain.
[0015] Figure 3 This application provides a schematic flowchart of a weather forecast optimization method based on complex terrain, which is used to obtain target features.
[0016] Figure 4 This application provides a schematic diagram of the structure of a weather forecast optimization system based on complex terrain.
[0017] Figure labeling: Historical data acquisition module 10, evaluation model building module 20, terrain influence factor acquisition module 30, weather forecast model building module 40, weather forecast information generation module 50. Detailed Implementation
[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0019] This application primarily addresses the problem that traditional weather forecasting methods neglect the specific impact of topography on weather, leading to significant errors, especially in areas with complex terrain. By analyzing a large amount of geographic information data and combining it with weather forecasting models, the accuracy of forecasts is improved.
[0020] To better understand the above technical solution, the following will provide a detailed description of the solution in conjunction with the accompanying drawings and specific implementation methods:
[0021] Example 1
[0022] like Figure 1 This paper presents a weather forecast optimization method based on complex terrain, the method comprising:
[0023] Using meteorological data acquisition equipment, collect topographic data and corresponding historical meteorological data within the target area;
[0024] Specifically, meteorological data acquisition equipment is used to collect topographic data and corresponding historical meteorological data within the target area to determine the target area: complex terrain refers to terrain with large undulations and diverse features, including mountains, plateaus, basins, hills, and plains. These terrains, due to their unique shapes and geographical locations, significantly impact weather and climate, thus affecting people's production and lives. For example, mountainous areas, due to their high altitude, have lower temperatures and relatively more precipitation; the direction and height of the mountains also influence airflow, leading to variable weather in mountainous regions. Plateaus, due to their high altitude, also have lower temperatures and relatively more precipitation; the longer hours of sunshine in plateau areas also influence weather and climate. Basins, due to their low-lying terrain, are prone to water accumulation, and the surrounding mountains also influence airflow, resulting in relatively stable weather in basin areas. Hilly areas have significant topographic undulations, affecting both temperature and precipitation; the high vegetation cover in hilly areas also impacts weather and climate. Plains have flat terrain, with relatively stable temperature and precipitation; however, the open views in plains can also have some influence on weather and climate. Selecting Meteorological Data Acquisition Equipment: Based on the characteristics of the target area and the types of data to be collected, select appropriate meteorological data acquisition equipment. This equipment may include automatic weather stations, weather balloons, satellite remote sensing devices, etc. Installing and Configuring Meteorological Data Acquisition Equipment: Install and configure the meteorological data acquisition equipment within the target area, ensuring that the equipment can correctly collect the required meteorological data. Simultaneously, these devices need to be connected to computers or other data collection systems for real-time data transmission and processing. Collecting Historical Meteorological Data: Use the meteorological data acquisition equipment to collect historical meteorological data corresponding to the topographic data. This data may include temperature, humidity, wind speed, rainfall, etc.
[0025] Establish a terrain impact assessment model;
[0026] Specifically, determine the model objective: clarify the objective and purpose of the model, such as evaluating the impact of terrain on weather factors such as rainfall, wind speed, air temperature, etc. Collect data: collect data related to terrain and weather, including terrain data, historical meteorological data, etc. Data preprocessing: clean, organize, and standardize the collected data to ensure the quality and consistency of the data. Feature extraction: extract relevant features from the data, such as terrain height, slope, aspect, etc., as well as temperature, humidity, wind speed, etc. in meteorological data. Model selection: select statistical models, neural networks, regression analysis, etc. according to the objective and data characteristics. Model training: use historical data to train the model to determine the model parameters. Model validation: use an independent data set to validate the model to ensure the accuracy and generalization ability of the model. Model application: apply the trained model to actual terrain data for terrain impact assessment and analysis.
[0027] Input the terrain data and the historical meteorological data into the terrain impact assessment model to obtain N terrain impact factors. The N terrain factors include M marker information, where M < N;
[0028] Specifically, inputting terrain data and historical meteorological data into the terrain impact assessment model can obtain N terrain impact factors. These factors usually include M marker information, where M < N. These factors can be obtained by analyzing the relationship between terrain data and historical meteorological data. For example, neural networks can be used to establish models, and terrain impact factors can be extracted from these models. Usually, the N terrain impact factors can include various factors such as terrain height, slope, aspect, vegetation coverage, geological structure, etc. These factors can have a direct or indirect impact on the weather, such as affecting rainfall, wind speed, air temperature, etc. After obtaining the N terrain impact factors, the M marker information can be determined by further analyzing the contribution degree of these factors. These marker information are areas with human habitation or construction.
[0029] Respectively use the N terrain impact factors as input data, and combine real-time meteorological data to construct a multi-level and multi-resolution weather forecasting model. The multi-level and multi-resolution weather forecasting model includes multiple sub-models, and each of the sub-models corresponds to one of the N terrain impact factors;
[0030] Specifically, by using N topographic influencing factors as input data and combining them with real-time meteorological data, a multi-level, multi-resolution weather forecasting model can be constructed. This model typically includes multiple sub-models, each corresponding to a specific topographic influencing factor. Data preparation: Collect and prepare input data, including topographic data, historical meteorological data, and real-time meteorological data. This data should be cleaned, organized, and standardized to ensure data quality and consistency. Model design: Design a multi-level, multi-resolution weather forecasting model based on the research objectives and data characteristics. This model typically includes multiple sub-models, each corresponding to a specific topographic influencing factor. Sub-model training: Train each sub-model using historical data to determine the parameters of each sub-model. Training can utilize methods such as regression analysis, neural networks, and support vector machines. Real-time data input: Input real-time meteorological data into the model as input data. This data can include real-time observational data such as temperature, humidity, wind speed, and air pressure. Model prediction: Use the trained sub-models to predict the real-time data to obtain future weather forecast results.
[0031] Weather forecast information is generated using the aforementioned weather forecast model.
[0032] Specifically, weather forecast information can be generated by constructing multi-layered, multi-resolution weather forecast models. Inputting Real-Time Data: Real-time observed meteorological data is input into the model as its input data. This data can include real-time observational data such as temperature, humidity, wind speed, and air pressure. Model Prediction: The trained sub-model is used to predict the real-time data to obtain future weather forecast results. Outputting Results: The prediction results are visualized in the form of graphs or charts for better understanding and presentation. These results can also be output to other applications or services, such as meteorological services, agricultural management, and urban planning. Generating Weather Forecast Information: Based on the prediction results and visualization results, weather forecast information is generated. This information can include predictions of weather conditions such as temperature, humidity, wind speed, and rainfall over a future period, as well as corresponding meteorological warnings and recommendations. Publishing Weather Forecast Information: The generated weather forecast information is published to relevant users or services. These users can include meteorological service agencies, agricultural management departments, and urban planning departments, while services can include television weather forecast programs, mobile applications, and websites.
[0033] Furthermore, such as Figure 2 As shown, the method of this application, the step of establishing a terrain impact assessment model, includes:
[0034] Obtain the topographic elevation map of the target area;
[0035] Based on the topographic elevation map, the target area is clustered according to its topographic features to obtain a topographic feature set;
[0036] Obtain the meteorological information corresponding to the terrain feature set;
[0037] The meteorological information of the terrain feature set is filtered according to a preset meteorological threshold to obtain a first terrain feature set, wherein the meteorological information of the first terrain feature set conforms to the preset meteorological threshold.
[0038] The target features are obtained, including meteorological time series data and a first set of terrain features, wherein the meteorological time series data and the first set of terrain features correspond one-to-one.
[0039] Based on the target characteristics, a terrain impact assessment model is constructed, wherein the target characteristics include meteorological time series data and a first set of terrain features.
[0040] Specifically, a topographic elevation map of the target area is obtained, and topographic features are clustered based on the elevation map to obtain a topographic feature set. These features may include elevation, slope, and aspect. Next, meteorological information corresponding to the topographic feature set is obtained, such as temperature, humidity, and wind speed. The meteorological information in the topographic feature set is filtered according to preset meteorological thresholds to obtain a first topographic feature set. The meteorological information in this first topographic feature set conforms to the preset meteorological thresholds, meaning the meteorological data is within a certain threshold range. Then, target features are obtained, which include meteorological time-series data and the first topographic feature set. The meteorological time-series data and the first topographic feature set are in a one-to-one correspondence. This meteorological time-series data may include meteorological data from a past period, such as data from the past week, month, or year. Finally, a topographic impact assessment model is constructed based on the target features. This model can be built using various machine learning algorithms or statistical methods, such as regression analysis, neural networks, and support vector machines. For example, when using regression analysis to build a topographic impact assessment model, it is first necessary to determine which topographic factors may affect the weather. These factors may include elevation, slope, vegetation type, and soil moisture. Use these factors as independent variables in the regression analysis. Clean, organize, and standardize the collected data. This may include handling missing values, outliers, and removing noise. Use appropriate regression methods (e.g., linear regression, logistic regression, decision tree regression, etc.) to build a model. Input the independent and dependent variables (weather indicators such as temperature, rainfall, etc.) into the model to build a terrain impact assessment model. Obtain a topographic elevation map of the target area: Topographic elevation maps of the target area can be obtained through various Geographic Information Systems (GIS) or remote sensing technologies. These elevation maps typically represent the degree of terrain undulation numerically and can reflect important terrain features. Perform terrain feature clustering based on the topographic elevation map: This step usually uses some data analysis and machine learning techniques, such as K-means clustering or DBSCAN clustering. For example, using DBSCAN clustering: First, the topographic elevation map needs to be converted into a data format suitable for DBSCAN processing. Elevation data can be considered as a feature, with the position and height of each pixel or data point as input data. Some data preprocessing may be required, such as removing outliers and standardizing data. The DBSCAN algorithm has two main parameters: Eps and MinPts. Eps defines the maximum distance between two data points, and MinPts defines the minimum number of neighbors around a data point. The choice of these two parameters directly affects the clustering results. Different parameter combinations can be tried to find the settings that best suit the data. Input the preprocessed data and the selected parameters into the DBSCAN algorithm to run the clustering process. DBSCAN returns a clustering result, where each cluster has a unique label.Visualizing clustering results provides a more intuitive understanding of the terrain distribution and clustering effectiveness. Color mapping can be used to represent different clusters with different colors, thus clearly displaying the clustering results of the terrain elevation map. The terrain data is then classified or clustered. This integrates similar terrain features together to form a terrain feature set. Meteorological information corresponding to the terrain feature set is then obtained: this information may come from weather stations, meteorological satellites, or other meteorological data sources. This data provides information on weather elements such as temperature, humidity, rainfall, and wind speed. The meteorological information in the terrain feature set is filtered according to preset meteorological thresholds: this step is to exclude data that does not meet the preset meteorological thresholds. For example, if the model only focuses on data with temperatures between 0-30 degrees Celsius, then data below or above this range will be excluded. Target features are obtained: these features include meteorological time-series data and the first terrain feature set. Meteorological time-series data provides meteorological information in the time dimension, while the first terrain feature set provides terrain feature information. These two types of data correspond one-to-one and together constitute the feature input of the model. Finally, the terrain impact assessment model is constructed based on the target features. This model could be a statistical model, a machine learning model, or a deep learning model. The model's input is the target features, and its output is an assessment of the impact of terrain on weather. Through this process, we can obtain a terrain impact assessment model that can evaluate the influence of terrain on weather. This type of model can help us better understand how terrain affects weather, thus playing an important role in fields such as weather forecasting and climate change research.
[0041] Furthermore, such as Figure 3 As shown, the method of this application obtains target features, and the method further includes:
[0042] Obtain meteorological time series data for the target area;
[0043] Dimensionality reduction techniques are used to perform feature dimensionality reduction on the meteorological time series data and the first terrain feature set to generate dimensionality-reduced features;
[0044] Establish a three-dimensional feature space;
[0045] The dimensionality-reduced features are mapped into the feature three-dimensional space to obtain the patterns and relationships of the feature data;
[0046] Based on the described patterns and relationships, feature cross-referencing is performed to obtain the target features.
[0047] Specifically, before constructing the terrain impact assessment model, it is very important to obtain the meteorological time series data of the target area. These data provide time series information about weather conditions and play a crucial role in evaluating the impact of terrain on weather. After obtaining the meteorological time series data, dimensionality reduction techniques can be used to reduce the dimensions of the data and the first terrain feature set to generate reduced-dimensional features. The dimensionality reduction techniques can include principal component analysis (PCA), linear discriminant analysis (LDA), t-SNE, etc. Through dimensionality reduction processing, high-dimensional data can be transformed into low-dimensional feature vectors, which is convenient for subsequent feature crossing and model training. After dimensionality reduction processing, a three-dimensional feature space can be established. This space can be used to represent the three-dimensional relationships between features and help us better understand and represent the patterns and relationships of the data. Specifically, this three-dimensional feature space can be a new feature space generated based on dimensionality reduction techniques such as PCA, LDA, or t-SNE. After mapping the reduced-dimensional features into the three-dimensional feature space, the patterns and relationships of the feature data can be obtained. These patterns and relationships can be used for feature crossing to obtain target features. Feature crossing refers to combining different features to generate new features to enhance the representation ability of the model. Through feature crossing, a richer feature set can be obtained, improving the prediction accuracy of the model.
[0048] Furthermore, in the method of this application, the terrain data and the historical meteorological data are input into the terrain impact assessment model to obtain N terrain impact factors. The N terrain factors include M marking information, where M < N. The method further includes:
[0049] Collect the user residence information and construction situation information in complex terrain areas;
[0050] Obtain M marking information;
[0051] Mark the areas with residence and construction based on the marking information;
[0052] If there is new construction information or new residents, update the marking.
[0053] Specifically, in areas with complex terrain, collecting information on resident populations and construction status is a crucial step, helping to understand the actual situation and make appropriate decisions. Collecting information on resident populations and construction status in areas with complex terrain includes surveys, mapping, and public data sources. The collected information should include, but is not limited to: the location, type, structure, and age of buildings; the type, start and end times of construction activities; and factors that may affect construction progress. Obtaining M marker information: This marker information is about terrain features, weather conditions, construction type and progress, resident type and distribution, etc. For example, a specific terrain feature may correspond to a specific resident type or a specific construction activity. Marking areas with residents and construction based on the marker information: This can be done through data analysis and visualization tools. For example, a Geographic Information System (GIS) can be used to mark and track changes in construction activities and resident distribution. Updating markers if new construction information or new residents are added: This step ensures the real-time nature and accuracy of the data. When new construction activities or new residents move in, the marker information should be updated to allow for appropriate adjustments to the model.
[0054] Furthermore, the method of this application also includes establishing a user feedback system and adjusting and optimizing the weather forecast model based on user feedback information.
[0055] Specifically, defining a user feedback system involves several steps: First, the goals and functions of the system need to be clearly defined. For example, it could collect user feedback on the accuracy of weather forecasts, their feelings about specific weather conditions, or suggestions for model functionality. The system can utilize multiple channels, such as a website, mobile application, or telephone hotline, to facilitate user feedback at any time. Collecting user feedback involves continuously collecting user feedback on the weather forecast model. This might include evaluations of model accuracy, feelings about specific weather conditions, and suggestions for model functionality. Analyzing user feedback requires analyzing and interpreting the collected feedback. Data analysis tools or manual analysis can be used to extract crucial information about model performance and user needs from the feedback. Adjusting and optimizing the weather forecast model involves identifying problems and shortcomings in the model based on the analyzed user feedback, and then making corresponding adjustments and optimizations. This might include improving the model's algorithm, adding new input data, and adjusting model parameters. By establishing a user feedback system, it's possible to ensure that the weather forecast model always aligns with user needs, improving the model's accuracy and usability. Simultaneously, it can enhance user trust and satisfaction with the model.
[0056] Furthermore, the method of this application also includes:
[0057] The first set of user feedback results is obtained through the user feedback system;
[0058] The accuracy of the first set of user feedback results is assessed.
[0059] If the judgment result is greater than the preset value, the second set of user feedback results is obtained by filtering.
[0060] Error analysis and confusion matrix were used to evaluate the performance of the weather forecast model and obtain the evaluation results.
[0061] The model parameters are adjusted and the model structure is improved based on the second set of user feedback results and the evaluation results.
[0062] Specifically, a first set of user feedback results can be obtained through the user feedback system. This set may include various user feedback on the weather forecast model, such as forecast accuracy and model response speed. Next, the accuracy of the first set of user feedback results is judged based on location and time markers. This is because weather variations can be significant in different regions and time periods, requiring differentiation of feedback results by region and time. If the judgment result is greater than a preset value, it indicates that the weather forecast model for that region performs well, and these feedback results can be further collected and analyzed to obtain a more comprehensive model performance evaluation. Otherwise, more data collection and analysis may be needed for that region to improve the model. Using error analysis, confusion matrices, and other methods to evaluate the performance of the weather forecast model is a crucial step. Error analysis helps us understand the distribution of the model's prediction errors, while confusion matrices show the model's performance across different categories. Based on the second set of user feedback results and the evaluation results, model parameters can be adjusted or the model structure improved. For example, if the model's prediction accuracy is found to be low in certain regions, model parameters can be adjusted or training data increased for those regions. If the overall performance of the model is good, but the prediction performance for certain categories is poor, the model structure can be improved to enhance the prediction performance for those categories. By obtaining user feedback on the weather forecast model through a user feedback system, and combining this feedback with location and time stamps to assess the accuracy of the results, and then adjusting and improving the model based on the assessment results, the performance of the weather forecast model can be gradually improved to better meet user needs.
[0063] Furthermore, the method of this application constructs a multi-level, multi-resolution weather forecast model, and the method further includes:
[0064] The spatiotemporal scope of weather forecasts is divided into different levels, with each level including different resolutions;
[0065] Different resolutions are defined across these multiple layers;
[0066] The target region is divided into descending resolution levels according to its size;
[0067] Constructing multi-level, multi-resolution weather forecasting models;
[0068] Weather forecast information is generated based on the weather forecast model.
[0069] Specifically, the spatiotemporal range of weather forecasts is divided into different levels: this can be done geographically, administratively, or in other meaningful ways, such as by province, city, county, township, etc. Each level should also include different resolutions to better meet the needs of different levels. Further subdivision of resolutions is possible across multiple levels: as mentioned above, each level can be further subdivided into different resolutions as needed. Here, we further divide the resolution into three levels: Level 1, Level 2, and Level 3. This division can be based on various factors, such as known meteorological data, required forecast accuracy, and the computational power of the model. Level 1 resolution is set for marked areas: for areas requiring special attention or already marked, we can set a higher resolution, i.e., Level 1. This allows for more accurate weather forecasts in these areas. Level 2 resolution is set for smaller areas: for areas with relatively small areas but still requiring high-accuracy forecasts, we can set a Level 2 resolution. This satisfies accuracy requirements while avoiding excessive computational burden. Setting a third-level resolution for large areas: For large areas, the focus may be more on general weather conditions rather than specific local details. Therefore, we can set a third-level resolution. This reduces computational load while maintaining necessary accuracy. Constructing a multi-level, multi-resolution weather forecast model: Based on the above settings, we can construct a multi-level, multi-resolution weather forecast model. This model should be able to generate weather forecast information at corresponding levels and resolutions based on different input data and parameters. Generating weather forecast information from the model: Finally, we can run the model, input relevant parameters and data, and generate weather forecast information at the corresponding levels and resolutions. This information can include, but is not limited to, temperature, humidity, wind speed, wind direction, and air pressure. In this way, we can achieve a weather forecast model that meets accuracy requirements while reducing computational burden. Furthermore, this method can be flexibly adjusted as needed. For example, during periods of high disaster incidence or special times, the resolution level of certain areas can be temporarily increased to provide more accurate weather forecasts.
[0070] Example 2
[0071] Based on the same inventive concept as the aforementioned embodiment of a weather forecast optimization method based on complex terrain, such as... Figure 4 As shown, this application provides a weather forecast optimization system based on complex terrain, the system comprising:
[0072] Historical data acquisition module 10 is used to collect topographic data and corresponding historical meteorological data within a target area using meteorological data acquisition equipment.
[0073] The evaluation model establishment module 20 is used to establish a terrain impact evaluation model.
[0074] The topographic impact factor acquisition module 30 is used to input the topographic data and the historical meteorological data into the topographic impact assessment model to acquire N topographic impact factors, wherein the N topographic factors include M label information, where M <N;
[0075] The weather forecast model building module 40 is used to take the N terrain influencing factors as input data and combine them with real-time meteorological data to build a multi-level, multi-resolution weather forecast model. The multi-level, multi-resolution weather forecast model includes multiple sub-models, and each sub-model corresponds one-to-one with the N terrain influencing factors.
[0076] Weather forecast information generation module 50, which is used to generate weather forecast information through the weather forecast model.
[0077] Furthermore, the system also includes:
[0078] A topographic impact assessment model construction module is used to: acquire a topographic elevation map of the target area; perform topographic feature clustering on the target area based on the topographic elevation map to obtain a topographic feature set; acquire meteorological information corresponding to the topographic feature set; filter the meteorological information of the topographic feature set according to a preset meteorological threshold to obtain a first topographic feature set, wherein the meteorological information of the first topographic feature set conforms to the preset meteorological threshold; acquire target features, wherein the target features include meteorological time series data and the first topographic feature set, wherein the meteorological time series data and the first topographic feature set correspond one-to-one; and construct a topographic impact assessment model based on the target features, wherein the target features include meteorological time series data and the first topographic feature set.
[0079] Furthermore, the system also includes:
[0080] The target feature acquisition module is used to obtain meteorological time series data of the target area; perform feature dimensionality reduction on the meteorological time series data and the first terrain feature set using dimensionality reduction technology to generate dimensionality-reduced features; establish a feature three-dimensional space; map the dimensionality-reduced features into the feature three-dimensional space to obtain the pattern and relationship of the feature data; and perform feature cross-interaction based on the pattern and relationship to obtain the target features.
[0081] Furthermore, the system also includes:
[0082] The area marking module is used to collect user residential information and construction information in areas with complex terrain; obtain M marking information; mark areas with residential areas and construction areas based on the marking information; and update the marking if there is new construction information or new residents.
[0083] Furthermore, the system also includes a feedback system establishment module for establishing a user feedback system and adjusting and optimizing the weather forecast model based on user feedback information.
[0084] Furthermore, the system also includes:
[0085] The model optimization module is used to obtain a first set of user feedback results through the user feedback system; to judge the accuracy of the first set of user feedback results; if the judgment result is greater than a preset value, to filter and obtain a second set of user feedback results; to use error analysis and confusion matrix to evaluate the performance of the weather forecast model and obtain evaluation results; and to adjust the model parameters and improve the model structure according to the second set of user feedback results and the evaluation results.
[0086] Furthermore, the system also includes:
[0087] The forecast information generation module is used to divide the spatiotemporal range of weather forecasts into different levels, wherein each level includes different resolutions; to divide different resolutions on the multiple levels; to divide the target area into descending resolution levels according to the size of the range; to construct a multi-level, multi-resolution weather forecast model; and to generate weather forecast information based on the weather forecast model.
[0088] Through the detailed description of the aforementioned weather forecast optimization method based on complex terrain, those skilled in the art can clearly understand the weather forecast optimization system based on complex terrain in this embodiment. As the system disclosed in the embodiment corresponds to the device disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.
[0089] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A weather forecast optimization method based on complex terrain, characterized in that, The method includes: Using a meteorological data collection device to collect topographic data within a target area and historical meteorological data corresponding to the topographic data; Establishing a topographic impact assessment model; Inputting the topographic data and the historical meteorological data into the topographic impact assessment model to obtain N topographic impact factors, where the N topographic factors include M marked information, and M < N; Using the N topographic impact factors as input data respectively, and combining real-time meteorological data to construct a multi-level and multi-resolution weather forecasting model, where the multi-level and multi-resolution weather forecasting model includes multiple sub-models, and each of the sub-models corresponds to one of the N topographic impact factors; Generating weather forecasting information through the weather forecasting model; The establishment of the topographic impact assessment model includes: Obtaining a topographic elevation map of the target area; Performing topographic feature clustering on the target area according to the topographic elevation map to obtain a set of topographic features; Obtaining meteorological information corresponding to the set of topographic features; Filtering the meteorological information of the set of topographic features according to a preset meteorological threshold to obtain a first set of topographic features, where the meteorological information of the first set of topographic features meets the preset meteorological threshold; Obtaining target features, where the target features include meteorological time series data and the first set of topographic features, and the meteorological time series data and the first set of topographic features correspond to each other; Constructing a topographic impact assessment model according to the target features, where the target features include meteorological time series data and the first set of topographic features.
2. The method as described in claim 1, characterized in that, Obtaining target features, and the method further includes: Obtaining meteorological time series data of the target area; Using a dimensionality reduction technique to perform feature dimensionality reduction on the meteorological time series data and the first set of topographic features to generate reduced-dimensional features; Establishing a feature three-dimensional space; Mapping the reduced-dimensional features into the feature three-dimensional space to obtain the patterns and relationships of the feature data; Performing feature crossing according to the patterns and relationships to obtain target features.
3. The method as described in claim 1, characterized in that, Inputting the topographic data and the historical meteorological data into the topographic impact assessment model to obtain N topographic impact factors, where the N topographic factors include M marked information, and M < N, and the method further includes: Collecting user residence information and construction situation information in complex terrain areas; Obtaining M marked information; Marking areas with residence and construction based on the marked information; Updating the marks if there is new construction information or new residents.
4. The method as described in claim 1, characterized in that, The method further includes establishing a user feedback system to adjust and optimize the weather forecasting model according to user feedback information.
5. The method as described in claim 4, characterized in that, It further includes: Obtaining a first set of user feedback results through the user feedback system; Judging the result accuracy of the first set of user feedback results; If the judgment result is greater than a preset value, screening to obtain a second set of user feedback results; Using error analysis and confusion matrix to evaluate the performance of the weather forecasting model to obtain an evaluation result; Adjusting model parameters and improving the model structure according to the second set of user feedback results and the evaluation result.
6. The method as described in claim 1, characterized in that, Constructing a multi-level and multi-resolution weather forecasting model, and the method further includes: The spatiotemporal scope of weather forecasts is divided into different levels, with each level including different resolutions; Different resolutions are defined across these multiple layers; The target region is divided into descending resolution levels according to its size; Constructing multi-level, multi-resolution weather forecasting models; Weather forecast information is generated based on the weather forecast model.
7. A weather forecast optimization system based on complex terrain, characterized in that, The system includes: The historical data acquisition module is used to collect topographic data and corresponding historical meteorological data within the target area using meteorological data acquisition equipment. An assessment model building module is used to build a terrain impact assessment model. The topographic impact factor acquisition module is used to input the topographic data and the historical meteorological data into the topographic impact assessment model to acquire N topographic impact factors, wherein the N topographic factors include M labeling information, where M... <N; A weather forecast model building module is used to take the N topographic influencing factors as input data and combine them with real-time meteorological data to build a multi-level, multi-resolution weather forecast model. The multi-level, multi-resolution weather forecast model includes multiple sub-models, and each sub-model corresponds one-to-one with the N topographic influencing factors. A weather forecast information generation module, which is used to generate weather forecast information through the weather forecast model; Furthermore, the system also includes: A topographic impact assessment model construction module is used to: acquire a topographic elevation map of the target area; perform topographic feature clustering on the target area based on the topographic elevation map to obtain a topographic feature set; acquire meteorological information corresponding to the topographic feature set; filter the meteorological information of the topographic feature set according to a preset meteorological threshold to obtain a first topographic feature set, wherein the meteorological information of the first topographic feature set conforms to the preset meteorological threshold; acquire target features, wherein the target features include meteorological time series data and the first topographic feature set, wherein the meteorological time series data and the first topographic feature set correspond one-to-one; and construct a topographic impact assessment model based on the target features, wherein the target features include meteorological time series data and the first topographic feature set.