Meteorological disaster early warning identification method, device and equipment for any point
By acquiring and analyzing real-time meteorological data, using multi-source data and numerical model forecasts, calculating and correcting meteorological disaster warnings, the problems of low warning accuracy caused by the lack of automatic sites in the existing technology and difficulty in fine-tuning manual analysis are solved, and the problem of low warning accuracy is achieved. Efficient meteorological disaster warnings at any point.
Patent Information
- Application Number
- CN202411007797.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-12-03
AI Technical Summary
The existing meteorological disaster warning technology has low accuracy in the absence of ground meteorological automatic stations, and it is difficult to make early warnings at a specific point due to manual analysis, resulting in disasters being unable to be discovered and warned in advance, resulting in loss of personnel and property.
By acquiring real-time meteorological data, using multi-source live fusion point analysis data and numerical mode forecast data with high spatial and temporal resolution, the spatial interpolation algorithm is used to obtain the observation analysis value and forecast data of any point, and calculate the between forecast data and real-time observation data. The deviation, statistical inspection indicators and average forecast errors are corrected for meteorological disaster warning identification.
It has realized the early warning and identification of meteorological disasters at any point, breaking the limitations of regional forecasts, improving the pertinence and accuracy of forecasts, reducing air reporting situations, and enhancing the disaster warning service capabilities at specific points.
Smart Images

Figure CN119065029A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of meteorological forecasting, and in particular to a method, device and equipment for early warning identification of meteorological disasters at any point. Background Art
[0002] With the global climate change and the frequent occurrence of extreme weather events, the accuracy and timeliness of meteorological disaster warnings are of great significance for protecting people's lives and property and guiding disaster prevention and mitigation work. Traditional meteorological disaster warning methods mainly rely on data from meteorological observation stations, as well as manual analysis and forecasters' experience and judgment. However, this method is limited by the density and distribution of observation stations, as well as the subjectivity and experience level of forecasters, resulting in certain limitations on the accuracy and efficiency of warnings.
[0003] In recent years, with the continuous advancement of meteorological observation technology and the rapid development of computer technology, meteorological disaster warning technology has also been significantly improved and upgraded. In particular, the integration of multi-source meteorological data, the development of numerical model forecasting technology, and the application of big data analysis technology have provided more comprehensive, accurate, and efficient data support and technical means for meteorological disaster warning.
[0004] At present, a large number of meteorological disaster warning technology research and practice work has been carried out at home and abroad. Among them, the fusion of multi-source meteorological data can make full use of the characteristics and advantages of various forecast data, improve the temporal and spatial resolution and accuracy of the data; numerical model forecasting technology can simulate and predict future weather conditions based on the principles of atmospheric dynamics and thermodynamics, and provide an important reference for meteorological disaster warning; big data analysis technology can discover the laws and trends of meteorological disasters through the mining and analysis of massive meteorological data, and provide a scientific basis for warning decisions.
[0005] However, there are still some problems and challenges in the current meteorological disaster warning technology. On the one hand, a considerable proportion of disaster-causing systems occur in places where ground meteorological automatic stations are sparsely distributed or even non-existent. In the absence of available automatic station observations, relying solely on numerical forecast products often results in the inability to detect and warn of disasters in advance, leading to unnecessary loss of life and property. On the other hand, when making disaster warning products, since human resources cannot distinguish whether each point on the map has the possibility of a disaster, they can only draw the range of possible disaster warnings in general, resulting in a large number of false reports in the affected area, resulting in low availability of the disaster affected area product when providing disaster warning services at a specific point.
[0006] Therefore, there is an urgent need to provide a more reliable meteorological disaster warning and identification solution. Summary of the invention
[0007] The purpose of the present invention is to provide a method, device and equipment for identifying meteorological disaster warnings at any point, which is used to solve the problem that the warning business in the prior art focuses more on regional forecasts, resulting in low accuracy of meteorological disaster warning identification results.
[0008] In order to achieve the above object, the present invention provides the following technical solutions:
[0009] In a first aspect, the present invention provides a method for identifying meteorological disaster warnings at any point, the method comprising:
[0010] Acquire real-time meteorological data; the meteorological data at least includes ground-based automatic meteorological station forecast data, multi-source live fusion grid analysis data, and numerical model real-time forecast data;
[0011] Based on multi-source real-time fusion grid analysis data with high temporal and spatial resolution, the observation analysis value of any point is obtained through spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of any point is obtained through spatial interpolation algorithm;
[0012] Based on the observed analysis values and the deviation between the forecast data and the actual observation data at each forecast time at each forecast starting time of the numerical model of each single-point forecast calculation;
[0013] Based on the forecast deviation data corresponding to each numerical mode at the forecast point obtained by the deviation calculation, the inspection index and the average forecast error of each forecast time of each numerical mode within the forecast time are counted;
[0014] Based on the verification index and the average forecast error during the training period of the target model, the forecast within the forecast period is corrected;
[0015] Meteorological disaster warning identification at any point based on revised forecasts.
[0016] Optionally, the correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period specifically includes:
[0017] Based on the accuracy of all the numerical modes of the inspection indicators, a ranking result is obtained;
[0018] determining a reference mode based on the sorting result;
[0019] The forecasts within the forecast period are corrected using the average forecast deviations during the reference model training period;
[0020] After correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period, the method further includes:
[0021] Based on the revised forecast of the reference model, meteorological disaster warning identification is carried out at any point, and warning identification products of the revised point forecast are intelligently recommended.
[0022] Optionally, the correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period specifically includes:
[0023] The average forecast deviation of each numerical model during the training period is used to correct the forecasts of each company within the forecast period;
[0024] After correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period, the method further includes:
[0025] Based on the revised forecasts of various companies, early warning identification of meteorological disasters at any point can be carried out, and early warning identification products of the revised point forecasts can be intelligently recommended.
[0026] Optionally, the multi-source real-time fusion grid analysis data with high spatiotemporal resolution is used to obtain the observation analysis value of any point through a spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of each company at any point is obtained through a spatial interpolation algorithm, specifically including:
[0027] Determine whether the position of the forecast point is consistent with the position of the ground meteorological automatic station, and obtain a first determination result;
[0028] When the first judgment result indicates that the location of the forecast point is consistent with the location of the ground automatic weather station, the weather forecast data of the current ground automatic weather station is determined as the forecast data of the forecast point;
[0029] When the first judgment result indicates that the position of the forecast point is inconsistent with the position of the ground automatic meteorological station, a first distance between the position of the forecast point and the nearest ground automatic meteorological station and a second distance between the position of the forecast point and the nearest grid point of the real-time grid analysis product are respectively calculated;
[0030] If the first distance is smaller than the second distance, the weather forecast data of the nearest ground weather automatic station is determined as the forecast data of the forecast point;
[0031] If the first distance is greater than the second distance, the meteorological forecast data of the grid point of the nearest real-time grid analysis product is determined as the forecast data of the forecast point.
[0032] Optionally, the multi-source real-time fusion grid analysis data with high spatiotemporal resolution is used to obtain the observation analysis value of any point through a spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of each company at any point is obtained through a spatial interpolation algorithm, specifically including:
[0033] Determine whether the forecast point position coincides with the numerical model grid point to obtain a second determination result;
[0034] When the second judgment result indicates that the position of the forecast point coincides with the numerical model grid point, determining the meteorological forecast data of the current numerical model grid point as the forecast data of the forecast point;
[0035] When the second judgment result indicates that the forecast point position does not overlap with the numerical model grid point, the meteorological forecast data of the numerical model grid point closest to the forecast point position is determined as the forecast data of the forecast point; or, spatial interpolation is performed based on the meteorological forecast data of multiple numerical model grid points within a preset distance range of the forecast point position to obtain the forecast data of the forecast point.
[0036] Optionally, the deviation between the forecast data of each forecast time period at each starting time based on the observation analysis value and the single-point forecast calculation numerical model of each company and the actual observation data specifically includes:
[0037] Using the formula:
[0038] B i =O i -F i
[0039] Calculate the deviation between the forecast data and the actual observation data at each forecast time of each forecast time of the numerical model; where i is the forecast time of each forecast time, O i represents the observation corresponding to the forecast time, F i is the forecast value corresponding to the forecast time, B i is the forecast bias of the forecast time effectiveness;
[0040] The method of correcting the forecast within the forecast period by using the average forecast deviation during the reference model training period specifically includes:
[0041] Based on the forecast deviation data and the reference model, the formula is adopted:
[0042]
[0043] Calculate the average of all forecast deviations at the start of each forecast period during the training period, where: is the average value of actual observations corresponding to the same forecast time during the training period, is the average forecast value of the same forecast time during the training period, is the average forecast deviation of each forecast time during the training period.
[0044] Optionally, the method of correcting the forecast within the forecast period using the average forecast deviation during the reference model training period specifically includes:
[0045] Using the average forecast bias during the reference model training period, the formula is:
[0046]
[0047] The forecast within the revised forecast period, where i represents the forecast time within the forecast period, BR i It is the revised forecast for each forecast period after the forecast point has been revised within the forecast period.
[0048] Compared with the prior art, the present invention provides a method for early warning and identification of meteorological disasters at any point, which obtains real-time meteorological data; obtains the observation analysis value of any point through a spatial interpolation algorithm based on multi-source live fusion grid analysis data with high temporal and spatial resolution; obtains the forecast data of each numerical model through a spatial interpolation algorithm for each single point forecast of any point; calculates the deviation between the forecast data of each forecast time period at each starting time of the numerical model and the actual observation data based on the forecast data; calculates the inspection index and average forecast error of each starting time forecast of each numerical model within the forecast time based on the forecast deviation data corresponding to the numerical model at the forecast point obtained by the deviation calculation; corrects the forecast within the forecast period based on the inspection index and the average forecast error during the training period of the target model; and performs early warning and identification of meteorological disasters at any point based on the corrected forecast. Breaking the process limitation of first making regional forecasts and then inverting them to the forecast points, directly selecting and correcting reference models for the forecast points makes the forecast more targeted. Any point in the country has the ability to make meteorological disaster warnings and identifications. The meteorological disaster warning identification process at any point can be expanded to multi-point identification, thereby realizing meteorological disaster warnings in any (irregular) area.
[0049] In a second aspect, the present invention provides a device for early warning and identifying meteorological disasters at any point, the device comprising:
[0050] A real-time meteorological data acquisition module is used to acquire real-time meteorological data; the meteorological data at least includes ground meteorological automatic station forecast data, multi-source real-time fusion grid analysis data and numerical model real-time forecast data;
[0051] The arbitrary point real-time observation module is used to obtain the observation analysis value of any point based on the multi-source real-time fusion grid analysis data with high temporal and spatial resolution through the spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of any point is obtained through the spatial interpolation algorithm;
[0052] A deviation calculation module, used for calculating the deviation between the forecast data and the actual observation data at each forecast time of each forecast time of the numerical model based on the observation analysis value and the single-point forecast of each company;
[0053] A verification index and average forecast error calculation module is used to calculate the forecast deviation data corresponding to each numerical mode at the forecast point obtained by the deviation calculation, and to calculate the verification index and average forecast error of each forecast time of each numerical mode within the forecast time;
[0054] A forecast correction module, used to correct the forecast within the forecast period based on the verification index and the average forecast error during the target model training period;
[0055] The meteorological disaster warning identification module is used to identify meteorological disaster warnings at any point based on the revised forecast.
[0056] In a third aspect, the present invention provides a meteorological disaster warning identification device at any point, the device comprising:
[0057] A memory, a processor and a communication interface coupled to the processor; the memory stores a computer program that can be run by the processor; when the processor runs the computer program, it executes the above-mentioned method for identifying meteorological disaster warnings at any point.
[0058] In a fourth aspect, the present invention provides a computer storage medium, wherein instructions are stored in the computer storage medium, and when the instructions are executed, the above-mentioned method for early warning and identifying meteorological disasters at any point is implemented.
[0059] The technical effects achieved by the device type solution provided in the second aspect, the equipment type solution provided in the third aspect, and the computer storage medium solution provided in the fourth aspect are the same as those of the method type solution provided in the first aspect, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS
[0060] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0061] Figure 1 A schematic diagram of a flow chart of a method for early warning and identifying meteorological disasters at any point provided by the present invention;
[0062] Figure 2 Schematic diagram for meteorological data collection;
[0063] Figure 3 It is a flowchart of the first method for extracting forecast data at any point;
[0064] Figure 4 This is a flow chart of the second method for extracting forecast data at any point;
[0065] Figure 5 Schematic diagram for calculating the average deviation of the pattern during the training period;
[0066] Figure 6 It is a reference forecast diagram during the revised forecast period;
[0067] Figure 7 This is a schematic diagram for meteorological disaster warning identification;
[0068] Figure 8 A complete flow chart of a method for early warning and identifying meteorological disasters at any point provided by the present invention;
[0069] Fig. 9 A schematic diagram of a meteorological disaster early warning identification device at any point provided by the present invention;
[0070] Fig.10 A schematic diagram of a meteorological disaster warning identification device for any point provided by the present invention. DETAILED DESCRIPTION
[0071] In order to clearly describe the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, words such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and effects. For example, the first threshold and the second threshold are only used to distinguish different thresholds, and their order is not limited. Those skilled in the art can understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit them to be different.
[0072] It should be noted that, in the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.
[0073] In the present invention, "at least one" means one or more, and "plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, the combination of a and b, the combination of a and c, the combination of b and c, or the combination of a, b and c, where a, b, c can be single or multiple.
[0074] In the prior art, when carrying out meteorological disaster warning, the usual steps are:
[0075] Obtain automatic station observation data such as precipitation, temperature, and wind.
[0076] Obtain grid data products such as numerical model forecasts.
[0077] On the electronic map of the meteorological visualization software, the actual and forecast data are superimposed and analyzed.
[0078] When actual observations of a certain area are superimposed on the forecast, or when conditions that meet a certain disaster threshold appear in the forecast, a corresponding closed curve will be drawn on the electronic map as the area where future disasters may occur.
[0079] The above-mentioned solutions in the prior art have the following problems:
[0080] 1) Data analysis is all done manually. At present, meteorological warning business products are mostly produced by forecasters who manually analyze the actual and numerical forecast data. With the increase in the types of meteorological numerical forecasts and the improvement in temporal and spatial resolution, it is difficult for humans to select the necessary information from the massive numerical forecast data within the specified time, thus increasing the possibility of missing potential disasters.
[0081] 2) The ground meteorological automatic station data used in the existing process cannot meet the needs of the development of high-precision early warning business. Due to the vast territory of my country, although the number of ground meteorological automatic observation stations in my country is about 70,000, the distribution of automatic stations is extremely uneven, with dense distribution in the central and eastern parts and sparse distribution in the west. Therefore, the distance between the two nearest automatic stations is also more than 5 kilometers, and the distance in areas with complex terrain can reach tens or hundreds of kilometers. Since the observation of ground meteorological automatic stations has a certain representative range, those located in flat terrain areas can represent the weather conditions within a radius of about 5-10 kilometers; in areas with complex terrain, the representative range can even be limited to the weather conditions within the station and a very small surrounding area. Against the background of global warming, disastrous weather occurs frequently in my country, and many of the disaster-causing weather systems are mostly small and medium-scale short-term severe convective systems, with diameters of about 1-50 kilometers. When these disaster-causing systems occur in places where ground meteorological automatic stations are sparsely distributed or even have no automatic stations, in the absence of available or reliable automatic station observations, relying solely on numerical forecast products often results in the inability to detect and warn disasters in advance, resulting in unnecessary losses of personnel and property.
[0082] 3) The existing warning business process and the products produced are not sophisticated enough. When producing disaster warning products, it is impossible to distinguish whether each point on the map has the possibility of disaster, so only the range of possible disaster warnings (which can be referred to as the disaster zone) can be roughly drawn, resulting in a large number of false alarms in the zone. As a result, when providing disaster warning services at a specific point, the availability of the disaster zone product is not high.
[0083] 4) Lack of business processes for meteorological disaster warning and identification at specific points. With the improvement of social economy, the business of meteorological disaster warning and identification at specific points has received more and more attention. Since the previous warning business focused more on disaster area forecasting, the meteorological industry currently lacks a set of mature point-based warning analysis and production processes.
[0084] In addition, before conducting meteorological disaster warning identification, the forecast needs to be corrected. In the existing technology, the test results of the mainstream global or regional numerical model forecasts commonly used in the business are usually retrieved through the test platform, and the recent numerical forecast results of various companies are compared and analyzed to select the best as a reference for the next forecast. Then, based on the forecaster's forecasting experience, the selected reference forecast is regionally corrected to achieve the purpose of being as accurate as possible under overall statistics. However, this correction method also has some problems, such as:
[0085] Manual analysis of test results cannot fully mine useful information. Daily forecast elements mainly include 2-meter temperature, precipitation, and 10-meter wind. The forecast of each element will generate multiple test index results such as root mean square error (RMSE), mean error (ME), and accuracy (HIT). In order to select a suitable reference forecast, it is necessary to conduct multi-angle comparative analysis between models and between several past reports of the model itself. Faced with a large amount of test data and analysis work, it is difficult to mine valuable information within a limited time. In actual work, especially when time is tight, the test result analysis link is often skipped, and the latest forecast of the commonly used model in the work is directly selected as the reference forecast to achieve the purpose of providing forecast results as soon as possible.
[0086] Moreover, the reference forecast selected by analyzing the regional forecast test results may not necessarily have the best forecast effect for the forecast point. At present, the test results of the forecast are mostly regional statistics, such as the statistics of the deviation between the observation and forecast of all ground meteorological automatic stations in the country or a province. The statistical results can represent the average situation of the forecast effect in the statistical area, but cannot accurately reflect the forecast effect of any point in the area, especially when the forecast point does not fall at the location of the ground meteorological automatic station. Therefore, the forecast selected by analyzing the total test statistics of all samples in the region has great uncertainty in the forecast effect of any single point in it.
[0087] In addition, the correction of the reference forecast lacks specificity. According to the existing single-point forecast business process, when correcting the reference forecast, the forecaster needs to face tens of thousands or even hundreds of thousands of forecast (grid) points in the forecast area. It is impossible to correct these points one by one by manpower, and only general regional corrections can be made, which will affect the final forecast accuracy of the forecast point.
[0088] At the same time, verification information is not fully utilized. Before calculating and counting all verification indicators, it is necessary to calculate the deviation between the actual observation and the forecast value within the forecast range, that is, the forecast deviation. This intermediate product of forecast verification is also the primary reference quantity in almost all primary (or manual) and objective forecast correction algorithms. However, since verification and forecast correction are currently divided into two business branches, the forecast deviation is repeatedly calculated and stored in a wasteful manner.
[0089] In order to solve the above-mentioned problems in the prior art, the present invention provides a method, device and equipment for meteorological disaster warning identification at any point. Specifically, in the technical solution provided by the present invention, on the one hand, based on the ground multi-source real-time fusion grid analysis data set, each numerical model forecast and the corresponding test results, the reference forecast selection and correction for any forecast point are carried out. On the other hand, based on the meteorological business products that have appeared in recent years, namely, the multi-source real-time fusion grid analysis data set, and the mainstream numerical model forecast with high spatiotemporal resolution, the meteorological warning identification for any point is designed.
[0090] Next, the solution provided by the embodiments of this specification is described in conjunction with the accompanying drawings:
[0091] like Figure 1 As shown, the process may include the following steps:
[0092] Step 110: Acquire real-time meteorological data; the meteorological data at least includes ground-based automatic meteorological station forecast data, multi-source live fusion grid analysis data, and numerical model real-time forecast data.
[0093] When this step is implemented, it can collect data on automatic station observations of surface meteorological elements (2-meter air temperature, 2-meter dew point temperature, 10-meter wind direction, 10-meter wind speed, surface air pressure, precipitation and visibility, etc.), multi-source real-time fusion grid analysis (referred to as real-time grid analysis data) and real-time forecasts of mainstream global and regional numerical models in China, such as Figure 1As shown in the figure, the real-time grid analysis data is based on the grid station data of the national ground meteorological automatic station observation, radar observation, satellite observation and numerical forecast, etc., and uses a two-dimensional variational algorithm to analyze the meteorological observation data grid data set of preset resolution spatial grid points, such as: 0.01° (1km) resolution spatial grid point (equal longitude and latitude) meteorological observation grid data set, the data accuracy is guaranteed. One of the advantages of this data set is that in areas where the ground meteorological automatic station is sparsely distributed or there is no automatic station, there are still reliable meteorological element analysis values that can be used as a reference and use.
[0094] Step 120: Based on the multi-source real-time fusion grid analysis data with high temporal and spatial resolution, the observation analysis value of any point is obtained through the spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of each point is obtained through the spatial interpolation algorithm.
[0095] High temporal and spatial resolution (e.g. 10-minute, 1km*1km grid). Each numerical model can represent the existing numerical models, such as: CMA_MESO, CMA_TYM and CMA_GFS developed by the National Meteorological Administration, CMA_SH, CMA_BJ and CMA_GD developed by the provincial level, and ECMWF and NCEP from abroad.
[0096] Of course, before implementing this step, you can set the threshold of any point of meteorological disasters. Types of meteorological disasters include heavy rain, heavy snow, cold wave, strong wind, high temperature, drought, frost, etc. Each disaster has a national meteorological disaster warning standard. In actual business, you can set a threshold standard suitable for the local area according to the local climate background of any point.
[0097] When real-time observation and forecast data are obtained at any point of a meteorological disaster, the meteorological observation data at the corresponding location can be selected as the forecast data for the arbitrary point based on the positional relationship between the arbitrary point, the ground automatic meteorological station and the grid points of the real-time grid analysis data.
[0098] Step 130: Based on the observed analysis values and the single-point forecasts of various companies, the deviation between the forecast data and the actual observed data at each forecast time of the numerical model is calculated.
[0099] Step 120 can be used to obtain the observation and forecast data of the forecast point, and calculate the deviation between the forecast data and the actual observation data for each forecast period at each start time of the numerical model. This verification step is automatically run at a fixed time every day, and the calculated forecast deviation data is stored for subsequent use.
[0100] Step 140: Based on the forecast deviation data corresponding to each numerical mode at the forecast point obtained by the deviation calculation, the verification index and the average forecast error of each forecast time of each numerical mode within the forecast time are statistically calculated.
[0101] Calculate each test index of each model at the forecast point within a period of time (training period). Then, based on the forecast deviation data of each model at the forecast point calculated in step 130, statistically calculate the test index of each model's forecast at each start time several days before the forecast time, such as root mean square error, average error or accuracy, etc. This statistical period is no less than 2 days, which is called the training period.
[0102] Each onset time has the same forecast validity period, such as 001h, 002h, ..., 024h, etc. Therefore, during the training period, the same forecast validity period will have multiple forecast values and forecast deviation data for different onset times.
[0103] Step 150: Based on the verification index and the average forecast error during the target model training period, the forecast within the forecast period is corrected.
[0104] The average forecast deviation of each forecast time during the training period is used to correct the reference forecast of each forecast time within the forecast period to obtain a more accurate forecast.
[0105] Specifically, based on the verification index and the average forecast error during the training period of the target model, the best short-term forecast effect can be selected from the numerical forecasts of various companies as a reference forecast and the forecast within the forecast period can be corrected, or based on the average error during the training period of each numerical model, the forecast within each forecast period can be corrected separately.
[0106] Step 160: Perform meteorological disaster warning identification at any point based on the revised forecast.
[0107] Based on the revised forecast, by calculating the actual observation data of any point and the forecast values of each numerical model, we can obtain the forecast results of each numerical model on the possibility of the occurrence of the meteorological disaster at that arbitrary point.
[0108] Figure 1The method in the invention obtains real-time meteorological data; obtains the observation analysis value of any point through the spatial interpolation algorithm based on the multi-source real-time fusion grid analysis data with high temporal and spatial resolution; obtains the single-point forecast of any point through the spatial interpolation algorithm based on the forecast data of each numerical model; calculates the deviation between the forecast data of each forecast time at each starting time of the numerical model and the actual observation data based on the forecast data; calculates the test index and average forecast error of each starting time of each numerical model within the forecast time based on the forecast deviation data corresponding to each numerical model at the forecast point obtained by the deviation calculation; corrects the forecast within the forecast period based on the test index and the average forecast error during the training period of the target model; and performs meteorological disaster warning identification at any point based on the corrected forecast. Breaking the process limitation of first making regional forecasts and then inverting them to the forecast points, directly selecting and correcting reference models for the forecast points makes the forecast more targeted. Any point in the country has the ability to make meteorological disaster warnings and identifications. The meteorological disaster warning identification process at any point can be expanded to multi-point identification, thereby realizing meteorological disaster warnings in any (irregular) area.
[0109] based on Figure 1 The method, the embodiments of this specification also provide some specific implementation methods of the method, which are described below.
[0110] Optionally, based on multi-source real-time fusion grid analysis data with high temporal and spatial resolution, the observation analysis value of any point is obtained through spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of any point is obtained through spatial interpolation algorithm, which specifically includes the following two implementation methods:
[0111] Method 1:
[0112] like Figure 3 As shown, the multi-source real-time fusion grid analysis data based on high temporal and spatial resolution is used to obtain the observation analysis value of any point through the spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of each point is obtained through the spatial interpolation algorithm, which may specifically include:
[0113] Determine whether the position of the forecast point is consistent with the position of the ground meteorological automatic station, and obtain a first determination result;
[0114] When the first judgment result indicates that the location of the forecast point is consistent with the location of the ground automatic weather station, the weather forecast data of the current ground automatic weather station is determined as the forecast data of the forecast point;
[0115] When the first judgment result indicates that the position of the forecast point is inconsistent with the position of the ground automatic meteorological station, a first distance between the position of the forecast point and the nearest ground automatic meteorological station and a second distance between the position of the forecast point and the nearest grid point of the real-time grid analysis product are respectively calculated;
[0116] If the first distance is smaller than the second distance, the weather forecast data of the nearest ground weather automatic station is determined as the forecast data of the forecast point;
[0117] If the first distance is greater than the second distance, the meteorological forecast data of the grid point of the nearest real-time grid analysis product is determined as the forecast data of the forecast point.
[0118] Method 2:
[0119] like Figure 4 As shown, the multi-source real-time fusion grid analysis data based on high temporal and spatial resolution is used to obtain the observation analysis value of any point through the spatial interpolation algorithm; based on the forecast data of each numerical model, the single-point forecast of each point is obtained through the spatial interpolation algorithm, which may specifically include:
[0120] Determine whether the forecast point position coincides with the numerical model grid point to obtain a second determination result;
[0121] When the second judgment result indicates that the position of the forecast point coincides with the numerical model grid point, determining the meteorological forecast data of the current numerical model grid point as the forecast data of the forecast point;
[0122] When the second judgment result indicates that the forecast point position does not coincide with the numerical model grid point, the meteorological forecast data of the numerical model grid point closest to the forecast point position is determined as the forecast data of the forecast point; or, spatial interpolation is performed based on the meteorological forecast data of multiple numerical (for example: 4 numerical) model grid points within a preset distance range of the forecast point position to obtain the forecast data of the forecast point, and using this method, the forecast values of each forecast time period of the numerical model are obtained in turn.
[0123] Optionally, in step 140, when calculating the forecast deviation of each model at the forecast point, first use step 120 to obtain the observation and forecast data of the forecast point, and calculate the deviation between the forecast data of each forecast time period at each start time of the numerical model and the actual observation data, as shown in formula (1). This verification step is automatically run at a scheduled time every day, and the calculated forecast deviation data is stored for subsequent use.
[0124] B i =O i -F i (1)
[0125] Among them, i is the forecast time validity under each starting time, O i is the observation corresponding to the forecast time, F i is the forecast value corresponding to the forecast time, B i It is the forecast deviation of forecast timeliness.
[0126] Optionally, based on the forecast deviation data corresponding to each numerical mode at the forecast point obtained by the deviation calculation, the verification index and the average forecast error of each forecast time of each numerical mode within the forecast time are counted, which may include:
[0127] Calculate the test indicators of each model at the forecast point within a period of time (training period). Based on the forecast deviation data of each model at the forecast point obtained by calculation, the test indicators of each model's forecast at each start time within a few days before the forecast time are statistically analyzed, such as root mean square error, average error or accuracy, as shown in formulas (2)-(4). This statistical period is no less than 2 days and is called the training period.
[0128]
[0129]
[0130] Where n is the total number of all forecast times at each onset time during the training period.
[0131] Optionally, based on the verification index and the average forecast error of the target model during the training period, the best model is selected as the reference model, and the forecast within the forecast period is corrected, or the forecasts within the forecast period of each model are corrected in turn. That is, based on the verification index and the average forecast error of the target model during the training period, the forecast within the forecast period is corrected. Specifically, two methods may be included:
[0132] Method 1:
[0133] Based on the accuracy of all the numerical modes of the inspection indicators, a ranking result is obtained;
[0134] determining a reference mode based on the sorting result;
[0135] The forecasts during the forecast period are corrected using the average forecast deviations during the reference model training period.
[0136] After correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period, the method may further include:
[0137] Based on the revised forecast of the reference model, meteorological disaster warning identification is carried out at any point, and warning identification products of the revised point forecast are intelligently recommended.
[0138] Method 2:
[0139] The average forecast deviation during the training period of each numerical model is used to correct the forecasts of each company within the forecast period.
[0140] After correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period, the method may further include:
[0141] Based on the revised forecasts of various companies, early warning identification of meteorological disasters at any point can be carried out, and early warning identification products of the revised point forecasts can be intelligently recommended.
[0142] In the first method, more specifically, the following methods can be used to implement it:
[0143] Select reference forecast: select one or more of the test indicators, such as minimum root mean square error, minimum average error, highest accuracy, or a combination of minimum root mean square error and highest accuracy, compare all the reporting times of all modes during the training period, and comprehensively select the forecast with the best forecast effect as the reference forecast. The mode corresponding to this forecast is called the reference mode.
[0144] Revised reference forecast: Figure 5 As shown in , the average forecast deviation of the reference model during the training period is calculated. Based on the calculated forecast deviation data, for the selected reference model, the average forecast deviation of all the starting times of each forecast time validity during the training period is calculated (Formula 5). Figure 5 , all the onset times during the training period are: 2 to N-1 onset times, each onset time has the same forecast time validity, such as 001h, 002h,..., 024h, etc. Therefore, during the training period, the same forecast time validity will have multiple forecast values and forecast deviation data of different onset times.
[0145]
[0146] in, is the average value of actual observations corresponding to the same forecast time during the training period, is the average forecast value of the same forecast time during the training period, is the average forecast deviation of each forecast time during the training period.
[0147] Revised reference forecast: Figure 6 As shown in FIG. 6 , the average forecast deviation of each forecast time in the training period is used to correct the reference forecast of each forecast time in the forecast period, as shown in formula (6), to obtain a more accurate forecast.
[0148]
[0149] Among them, BR i It is the revised forecast for each forecast period after the forecast point has been revised within the forecast period.
[0150] The steps included in the above method 1 are the process of selecting and correcting the reference forecast for any forecast point, which can be organized into the following steps:
[0151] Step 1: Collect ground-based automatic station observation data, multi-source real-time fusion grid analysis data, and forecasts from various numerical models.
[0152] Step 2: Extract or generate actual observation and forecast data for the points that need forecasting
[0153] Step 3: Calculate the deviation between the actual situation and the forecast at the forecast point in real time and store the relevant data.
[0154] Step 4: According to the forecast time of the forecast point, statistically calculate the test indicators of the numerical model forecasts of various companies within a few days before the forecast time (training period), and select the appropriate reference model and corresponding reference forecast according to certain standards.
[0155] Step 5: Use the average forecast deviation of the reference model during the training period to correct the forecast values of each forecast period of the reference forecast during the forecast period to obtain the revised forecast for the forecast point.
[0156] Step 6: Repeat steps 2-5 to obtain a regional forecast consisting of several points.
[0157] Repeat the above steps for each forecast point in a forecast area to generate a forecast for the area.
[0158] Through the above correction steps, we can break the process limitation of first making regional forecasts and then inverting to forecast points, directly select and correct reference models for forecast points, make forecasts more targeted, and help improve forecast accuracy. At the same time, using this process, we can make full use of the intermediate products and statistical results of the inspection for the selection and correction of reference forecasts, saving computing power and storage resources. Furthermore, by transforming the process into machine language, it can be used to make forecasts for each forecast point in a specific area, thereby generating a regional forecast.
[0159] Optional, the meteorological disaster warning identification process at any point is as follows Figure 7 As shown, the following steps may be included:
[0160] Obtain pre-set meteorological disaster warning threshold standards;
[0161] By using the threshold standard, we can judge whether the actual observation data of any point and the forecast data of any point, as well as the sum of the actual observation and forecast data of any point, reach or exceed the threshold, and then get the possibility of the meteorological disaster currently occurring at any point, or about to occur or in the future; repeat this step, combine the actual observation data of any point with the forecast value of each numerical model, and get the forecast result of each numerical model on the possibility of the meteorological disaster occurring at the arbitrary point. For example: if the actual observation data of the early warning point is greater than or equal to the threshold, it is determined that the disaster has occurred in the actual situation; if the alarm data of the early warning point is greater than or equal to the threshold, the disaster will occur in the future; if the sum of the actual observation data and forecast data of the early warning point is greater than or equal to the threshold, it is determined that the disaster is about to occur.
[0162] Such results can be presented in a simple and clear "yes / no" form. By comparing and analyzing these results, more accurate conclusions can be drawn based on full consideration of the forecast information of each model.
[0163] If it is necessary to carry out meteorological disaster early warning identification for a certain area, it is only necessary to repeat the above steps for each point in the area to obtain the results of meteorological disaster identification by various numerical models in the area.
[0164] It should be noted that any point in the present invention may be a forecast point or a warning point. In this specification, corresponding expressions are used in different scenarios.
[0165] In the above-mentioned meteorological disaster warning identification step, the ground-based real-time automatic station observation data, multi-source real-time fusion grid analysis data and forecasts of various numerical models required for meteorological disaster warning identification are obtained. According to the climate background of the location of any point where meteorological disaster warning identification is required, a suitable meteorological disaster threshold standard is set, and the ground meteorological automatic station observation and the ground multi-source meteorological real-time grid analysis product set are used to produce the real-time observation data of any point. The forecast data of any point is produced using the numerical model forecast. When there are multiple numerical models, the forecast data of each model at any point is produced separately. Meteorological disaster warning identification uses the set meteorological disaster threshold to filter the real-time observation data, forecast data and the sum of real-time observation and forecast data of any point, and then obtain the warning identification results of whether meteorological disasters occur in the current, future and near periods. Breaking the limitation that the traditional meteorological disaster warning identification business only has disaster warning products, any point in the country has the ability to produce meteorological disaster warning identification. Through the meteorological disaster warning identification process at any point, it can be expanded to multi-point identification, thereby realizing meteorological disaster warning in any (irregular) area.
[0166] Further, the complete technical solution of this application is combined with Figure 8 To explain:
[0167] First, the forecasts of various numerical models of the actual observation grid station data are obtained, and the actual situation and forecast data of each numerical model at any point are produced. The forecast deviation of each numerical model forecast at the point is calculated in real time, and the test index and average forecast deviation of each numerical model at the point during the training period are counted in real time. After that, on the one hand, the optimal forecast can be directly corrected: the forecast within the forecast period is corrected using the average forecast deviation of the reference model during the training period, and then the meteorological disaster warning identification at any point is carried out, and the warning identification product of the corrected point forecast is intelligently recommended; on the other hand, after obtaining the test index and the average forecast deviation, all numerical models can be corrected according to the test results. Accuracy sorting and intelligent recommendation: select the best model as the reference model, and then use the average forecast deviation of the reference model during the training period to correct the forecast within the forecast period, perform early warning identification of meteorological disasters at any point, and intelligently recommend the early warning identification products of the corrected point forecast; thirdly, after obtaining the test index and the average forecast deviation, all forecasts can be corrected: use the average forecast deviation of each numerical model during the training period to correct the forecast of each company within the forecast period, and then perform early warning identification of meteorological disasters at any point, and obtain the early warning identification products of each model forecast at that point, and calculate the probability of disaster occurrence based on the early warning identification results of the corrected forecasts of each model.
[0168] The technical solution provided by the present invention provides a feasible solution for meteorological disaster warning identification at any point in the country. When making regional disaster warning products, compared with existing disaster warning products, this method can effectively eliminate points in the region with low or no disaster possibility, thereby reducing false reports. By transforming the process into machine language and forming an automatically running system, the forecast results of various numerical models can be automatically analyzed. While saving manpower, the advantages of massive numerical forecast data can be fully utilized to obtain more comprehensive meteorological disaster warning identification results, providing a basis for making more accurate warning products.
[0169] It should be noted that the present invention includes two key points: intelligent recommendation and meteorological warning identification. When making intelligent recommendations, the average forecast deviation during the reference model training period can be used to correct the forecast within the forecast period, and then any point meteorological disaster warning identification can be performed, and the warning identification product of the corrected point forecast can be intelligently recommended; another way is to obtain the test index and the average forecast deviation, and then sort the accuracy of all numerical models according to the test results, and make intelligent recommendations: select the optimal model as the reference model, and then use the average forecast deviation during the reference model training period to correct the forecast within the forecast period, and then any point meteorological disaster warning identification can be performed, and the warning identification product of the corrected point forecast can be intelligently recommended.
[0170] Based on the same idea, the present invention also provides a meteorological disaster warning identification device at any point, such as Fig. 9As shown, the device may include:
[0171] A real-time meteorological data acquisition module 910 is used to acquire real-time meteorological data; the meteorological data at least includes ground meteorological automatic station forecast data, multi-source live fusion grid analysis data and numerical model real-time forecast data;
[0172] The arbitrary point live observation module 920 is used to obtain the observation analysis value of any point through the spatial interpolation algorithm based on the multi-source live fusion grid analysis data with high temporal and spatial resolution; and obtain the single point forecast of any point through the spatial interpolation algorithm based on the forecast data of each numerical model;
[0173] Deviation calculation module 930, used for calculating the deviation between the forecast data and the actual observation data at each forecast time of each forecast time in the numerical model based on the observation analysis value and each single point forecast;
[0174] The verification index and average forecast error calculation module 940 is used to calculate the verification index and average forecast error of each forecast of each numerical mode within the forecast time based on the forecast deviation data corresponding to each numerical mode at the forecast point obtained by the deviation calculation;
[0175] A forecast correction module 950, configured to correct the forecast within the forecast period based on the verification index and the average forecast error during the target model training period;
[0176] The meteorological disaster warning identification module 960 is used to identify meteorological disaster warnings at any point based on the revised forecast.
[0177] based on Fig. 9 The device in the embodiment may further include some specific implementation units:
[0178] Optionally, the forecast correction module 950 may specifically include:
[0179] A sorting unit, used to sort the accuracy of all numerical modes of the inspection indicators to obtain a sorting result;
[0180] a reference mode determining unit, configured to determine a reference mode based on the sorting result;
[0181] The first forecast correction unit is used to correct the forecast within the forecast period using the average forecast deviation during the reference model training period.
[0182] The device may further include:
[0183] The first intelligent recommendation module is used to identify meteorological disaster warnings at any point based on the forecast revised by the reference model, and intelligently recommend warning identification products of the revised point forecast.
[0184] Optionally, the forecast correction module 950 may specifically include:
[0185] The second forecast correction unit is used to correct the forecasts of various companies within the forecast period using the average forecast deviation during the training period of each numerical model.
[0186] The device may further include:
[0187] The second intelligent recommendation module is used to identify meteorological disaster warnings at any point based on the revised forecasts of various companies, and intelligently recommend warning identification products for the revised point forecasts.
[0188] Optionally, the arbitrary point live observation module 920 may specifically include:
[0189] A first judgment unit is used to judge whether the position of the forecast point is consistent with the position of the ground meteorological automatic station, and obtain a first judgment result;
[0190] A first forecast data determination unit, configured to determine the weather forecast data of the current ground weather automatic station as the forecast data of the forecast point when the first judgment result indicates that the location of the forecast point is consistent with the location of the ground weather automatic station;
[0191] a distance calculation unit, for calculating, when the first judgment result indicates that the position of the forecast point is inconsistent with the position of the ground automatic meteorological station, a first distance between the position of the forecast point and the nearest ground automatic meteorological station and a second distance between the position of the forecast point and the nearest grid point of the real-time grid analysis product;
[0192] a second forecast data determination unit, configured to determine the weather forecast data of the nearest ground weather automatic station as the forecast data of the forecast point if the first distance is smaller than the second distance;
[0193] The third forecast data determination unit is used to determine the meteorological forecast data of the nearest real-time grid analysis product grid point as the forecast data of the forecast point if the first distance is greater than the second distance.
[0194] Optionally, the arbitrary point live observation module 920 may specifically include:
[0195] A second judgment unit is used to judge whether the forecast point position coincides with the numerical model grid point, and obtain a second judgment result;
[0196] a third forecast data determination unit, configured to determine the meteorological forecast data of the current numerical model grid point as the forecast data of the forecast point when the second judgment result indicates that the position of the forecast point coincides with the numerical model grid point;
[0197] The fourth determination unit of forecast data is used to determine the meteorological forecast data of the numerical model grid point closest to the position of the forecast point as the forecast data of the forecast point when the second judgment result indicates that the position of the forecast point does not coincide with the numerical model grid point; or to obtain the forecast data of the forecast point by performing spatial interpolation based on the meteorological forecast data of multiple numerical model grid points within a preset distance range of the forecast point position.
[0198] Optionally, the deviation calculation module 930 may be specifically used for:
[0199] Using the formula:
[0200] B i =O i -F i
[0201] Calculate the deviation between the forecast data and the actual observation data at each forecast time of each forecast time of the numerical model; where i is the forecast time of each forecast time, O i represents the observation corresponding to the forecast time, F i is the forecast value corresponding to the forecast time, B i is the forecast bias of the forecast time effectiveness;
[0202] The forecast correction module 950 can be specifically used for:
[0203] Based on the forecast deviation data and the reference model, the formula is adopted:
[0204]
[0205] Calculate the average of all forecast deviations at the start of each forecast period during the training period, where: is the average value of actual observations corresponding to the same forecast time during the training period, is the average forecast value of the same forecast time during the training period, is the average forecast deviation of each forecast time during the training period.
[0206] Optionally, the forecast correction module 950 may be specifically used for:
[0207] Using the average forecast bias during the reference model training period, the formula is:
[0208]
[0209] The forecast within the revised forecast period, where i represents the forecast time within the forecast period, BR i It is the revised forecast for each forecast period after the forecast point has been revised within the forecast period.
[0210] Based on the same idea, the embodiment of this specification also provides a meteorological disaster warning identification device at any point. Fig.10 As shown, the device includes:
[0211] A memory, a processor and a communication interface coupled to the processor; the memory stores a computer program that can be run by the processor; when the processor runs the computer program, it executes the aforementioned method for identifying meteorological disaster warnings at any point.
[0212] like Fig.10 As shown, the processor may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present invention. The communication interface may be one or more. The communication interface may use any transceiver or other device for communicating with other devices or communication networks.
[0213] like Fig.10 As shown, the terminal device may further include a communication line. The communication line may include a path to transmit information between the components.
[0214] Optional, such as Fig.10 As shown, the terminal device may further include a memory. The memory stores a computer program that can be executed by the processor; when the processor executes the computer program, the method provided by the embodiment of the present invention is implemented.
[0215] like Fig.10As shown, the memory may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may exist independently and be connected to the processor through a communication line. The memory may also be integrated with the processor.
[0216] Optionally, the computer-executable instructions in the embodiment of the present invention may also be referred to as application program codes, which is not specifically limited in the embodiment of the present invention.
[0217] In a specific implementation, as an example, Fig.10 As shown, the processor may include one or more CPUs, such as Fig.10 CPU0 and CPU1 in.
[0218] In a specific implementation, as an example, Fig.10 As shown, the terminal device may include multiple processors, such as Fig.10 Each of these processors can be a single-core processor or a multi-core processor.
[0219] Based on the same idea, the embodiments of this specification also provide a computer storage medium corresponding to the above embodiments, and the computer storage medium stores instructions, and when the instructions are executed, the method in the above embodiments is implemented.
[0220] The above mainly introduces the solution provided by the embodiment of the present invention from the perspective of the interaction between the various modules. It is understandable that, in order to realize the above functions, each module includes a hardware structure and / or software unit corresponding to the execution of each function. It should be easily appreciated by those skilled in the art that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present invention can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.
[0221] The embodiment of the present invention can divide the functional modules according to the above method example. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation.
[0222] The processor in this specification may also have the function of a memory. The memory is used to store computer-executable instructions for executing the solution of the present invention, and the execution is controlled by the processor. The processor is used to execute the computer-executable instructions stored in the memory, thereby implementing the method provided by the embodiment of the present invention.
[0223] The memory may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may exist independently and be connected to the processor through a communication line. The memory may also be integrated with the processor.
[0224] Optionally, the computer-executable instructions in the embodiment of the present invention may also be referred to as application program codes, which is not specifically limited in the embodiment of the present invention.
[0225] The method disclosed in the above embodiment of the present invention can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor may be a general-purpose processor, a digital signal processor (digital signal processing, DSP), an ASIC, a field-programmable gate array (field-programmable gate array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.
[0226] Although the present invention is described herein in conjunction with various embodiments, in the process of implementing the claimed invention, those skilled in the art may understand and implement other variations of the disclosed embodiments by viewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "one" or "an" does not exclude multiple situations. A single processor or other unit may implement several functions listed in the claims. Certain measures are recorded in different dependent claims, but this does not mean that these measures cannot be combined to produce good results.
[0227] Although the present invention has been described in conjunction with specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present invention. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present invention as defined by the appended claims and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present invention. Obviously, those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, the present invention is intended to include such modifications and variations if they fall within the scope of the claims of the present invention and their equivalents.
Claims
1. A method for early warning and identification of meteorological disasters at any point, characterized in that: Methods include: Get real-time weather data; The meteorological data at least includes ground-based automatic meteorological station forecast data, multi-source real-time fusion grid analysis data, and numerical model real-time forecast data; Based on multi-source real-time fusion grid analysis data with high temporal and spatial resolution, the observation analysis value of any point is obtained through spatial interpolation algorithm; Based on the forecast data of each numerical model, the single point forecast of each company at any point is obtained through spatial interpolation algorithm; Based on the observed analysis values and the deviation between the forecast data and the actual observation data at each forecast time at each forecast starting time of the numerical model of each single-point forecast calculation; Based on the forecast deviation data corresponding to each numerical mode at the forecast point obtained by the deviation calculation, the inspection index and the average forecast error of each forecast time of each numerical mode within the forecast time are counted; Based on the verification index and the average forecast error during the training period of the target model, the forecast within the forecast period is corrected; Meteorological disaster warning identification at any point based on revised forecasts.
2. The method for early warning and identification of meteorological disasters at any point according to claim 1, characterized in that: The correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period specifically includes: Based on the accuracy of all the numerical modes of the inspection indicators, a ranking result is obtained; determining a reference mode based on the sorting result; The forecasts within the forecast period are corrected using the average forecast deviations during the reference model training period; After correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period, the method further includes: Based on the revised forecast of the reference model, meteorological disaster warning identification is carried out at any point, and warning identification products of the revised point forecast are intelligently recommended.
3. The method for early warning and identification of meteorological disasters at any point according to claim 1, characterized in that: The correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period specifically includes: The average forecast deviation of each numerical model during the training period is used to correct the forecasts of each company within the forecast period; After correcting the forecast within the forecast period based on the verification index and the average forecast error during the target model training period, the method further includes: Based on the revised forecasts of various companies, early warning identification of meteorological disasters at any point can be carried out, and early warning identification products of the revised point forecasts can be intelligently recommended.
4. The method for early warning and identification of meteorological disasters at any point according to claim 1, characterized in that: The multi-source real-time fusion grid analysis data based on high temporal and spatial resolution is used to obtain the observation analysis value of any point through a spatial interpolation algorithm; Based on the forecast data of each numerical model, the single-point forecast of any point is obtained through the spatial interpolation algorithm, including: Determine whether the position of the forecast point is consistent with the position of the ground meteorological automatic station, and obtain a first determination result; When the first judgment result indicates that the location of the forecast point is consistent with the location of the ground automatic weather station, the weather forecast data of the current ground automatic weather station is determined as the forecast data of the forecast point; When the first judgment result indicates that the position of the forecast point is inconsistent with the position of the ground meteorological automatic station, a first distance between the position of the forecast point and the nearest ground meteorological automatic station and a second distance between the position of the forecast point and the nearest grid point of the real-time grid analysis product are respectively calculated; If the first distance is smaller than the second distance, the weather forecast data of the nearest ground weather automatic station is determined as the forecast data of the forecast point; If the first distance is greater than the second distance, the meteorological forecast data of the grid point of the nearest real-time grid analysis product is determined as the forecast data of the forecast point.
5. The method for early warning and identification of meteorological disasters at any point according to claim 1, characterized in that: The multi-source real-time fusion grid analysis data based on high temporal and spatial resolution is used to obtain the observation analysis value of any point through a spatial interpolation algorithm; Based on the forecast data of each numerical model, the single-point forecast of any point is obtained through the spatial interpolation algorithm, including: Determine whether the forecast point position coincides with the numerical model grid point to obtain a second determination result; When the second judgment result indicates that the position of the forecast point coincides with the numerical model grid point, determining the meteorological forecast data of the current numerical model grid point as the forecast data of the forecast point; When the second judgment result indicates that the forecast point position does not overlap with the numerical model grid point, the meteorological forecast data of the numerical model grid point closest to the forecast point position is determined as the forecast data of the forecast point; or, spatial interpolation is performed based on the meteorological forecast data of multiple numerical model grid points within a preset distance range of the forecast point position to obtain the forecast data of the forecast point.
6. The method for early warning and identification of meteorological disasters at any point according to claim 2, characterized in that: The deviation between the forecast data of each forecast time period at each forecast starting time based on the observation analysis value and the single-point forecast calculation numerical model of each company and the actual observation data specifically includes: Using the formula: B i =O i -F i Calculate the deviation between the forecast data and the actual observation data at each forecast time of each forecast time of the numerical model; where i is the forecast time of each forecast time, O i represents the observation corresponding to the forecast time, F i is the forecast value corresponding to the forecast time, B i is the forecast bias of the forecast time effectiveness; The method of correcting the forecast within the forecast period by using the average forecast deviation during the reference model training period specifically includes: Based on the forecast deviation data and the reference model, the formula is adopted: Calculate the average of all forecast deviations at the start of each forecast period during the training period, where: is the average value of actual observations corresponding to the same forecast time during the training period, is the average forecast value of the same forecast time during the training period, is the average forecast deviation of each forecast time during the training period.
7. The method for early warning and identification of meteorological disasters at any point according to claim 6, characterized in that: The method of correcting the forecast within the forecast period by using the average forecast deviation during the reference model training period specifically includes: Using the average forecast bias during the reference model training period, the formula is: The forecast within the revised forecast period, where i represents the forecast time within the forecast period, BR i It is the revised forecast for each forecast period after the forecast point has been revised within the forecast period.
8. A meteorological disaster warning identification device at any point, characterized in that: The device includes: Real-time meteorological data acquisition module, used to obtain real-time meteorological data; The meteorological data at least includes ground-based automatic meteorological station forecast data, multi-source real-time fusion grid analysis data, and numerical model real-time forecast data; The arbitrary point real-time observation module is used to obtain the observation analysis value of any point through the spatial interpolation algorithm based on multi-source real-time fusion grid analysis data with high temporal and spatial resolution; Based on the forecast data of each numerical model, the single point forecast of each company at any point is obtained through spatial interpolation algorithm; A deviation calculation module, used for calculating the deviation between the forecast data and the actual observation data at each forecast time of each forecast time of the numerical model based on the observation analysis value and the single-point forecast of each company; A verification index and average forecast error calculation module is used to calculate the forecast deviation data corresponding to each numerical mode at the forecast point obtained by the deviation calculation, and to calculate the verification index and average forecast error of each forecast time of each numerical mode within the forecast time; A forecast correction module, used to correct the forecast within the forecast period based on the verification index and the average forecast error during the target model training period; The meteorological disaster warning identification module is used to identify meteorological disaster warnings at any point based on the revised forecast.
9. A meteorological disaster warning identification device at any point, characterized in that the device include: A memory, a processor, and a communication interface coupled to the processor; The memory stores a computer program executable by the processor; When the processor runs the computer program, it executes the meteorological disaster warning identification method for any point as described in any one of claims 1 to 7.
10. A computer storage medium, characterized in that: The computer storage medium stores instructions, and when the instructions are executed, the method for early warning and identification of meteorological disasters at any point described in any one of claims 1 to 7 is implemented.
Citation Information
Cited By
A highway weather grading early warning method, system, device and medium
CN122888814A
