Objective forecasting method and device for 100-meter level gusts

Abstract

This invention relates to a method and apparatus for objective gust forecasting at a resolution of 100 meters. The method includes acquiring weather element data; performing quantitative analysis on multiple preset intervals to obtain a gust coefficient model; the preset intervals include intervals of different altitudes, different wind speeds, and different wind directions; interpolating the gust coefficient model into a grid field with a resolution of 100 meters to obtain a gust coefficient grid field; obtaining a preset average wind based on a pre-acquired average wind point deviation correction coefficient at a resolution of 100 meters, determining the range of the gust coefficient interval and its grid field, and obtaining gust point prediction data at a resolution of 100 meters. This invention obtains the predicted gust by constructing a gust coefficient model and combining it with the average wind point deviation correction coefficient at a resolution of 100 meters to obtain the corrected average wind of the Ruisi system. This application can improve the level of refined gust forecasting under complex terrain conditions and enhance the safety operation and disaster prevention and mitigation capabilities of major events and cities in the capital and surrounding areas.

Description

Technical Field

[0001] This invention belongs to the field of meteorological forecasting technology, specifically relating to a method and device for objective forecasting of gusts at the 100-meter level. Background Technology

[0002] Most of northern my country, especially the Beijing-Tianjin-Hebei region, is influenced by continental monsoon climates. Cold air originates from high-latitude continental areas, primarily blowing northerly and northwesterly winds. When strong cold air masses invade from the south, they often cause drastic temperature drops, accompanied by non-convective strong winds. This can significantly impact energy production, aviation, transportation, power equipment, the environment, agriculture, and construction. Furthermore, people tend to pay less attention to high-impact non-convective winds compared to tornadoes or thunderstorms, meaning they can potentially cause more damage and casualties than thunderstorms or hurricanes. In addition, through participation in meteorological support for scientific research and services, we have gained a deeper understanding of the challenges of small-scale mountain wind forecasting in winter, especially gust forecasting, and the importance of precise monitoring, accurate forecasting, and refined services for major outdoor events under different terrain conditions. Currently, reliable methods for gust forecasting are still lacking. Numerical weather prediction models can rely on measurements of wind speed and turbulence to forecast gusts, but their level of detail cannot reflect the dynamic and nonlinear flow characteristics of the wind field at the microscale. The forecast accuracy is also relatively low in complex terrain areas, failing to meet the service needs for precise forecasts.

[0003] Among related technologies, one of the main methods for gust forecasting is the physical model based on boundary layer turbulence theory. However, this parameterization scheme only effectively simulates terrain drag under synoptic scale systems. In complex terrain, many stations are located in mountainous areas with numerous narrow fjords and valleys, and the dynamic and nonlinear flow characteristics of the wind field at the microscale can have a significant impact on local winds. Another statistical forecasting method based on gust factors combines local climate measurements of gusts with wind speed forecasts. However, the gust factor model only couples the station coefficients with the model station outputs and does not integrate the gust factor model into the high-resolution model. Therefore, it cannot obtain high-precision gridded gust forecasts and still cannot meet the rigid demand of the major meteorological support services in the Beijing-Tianjin-Hebei region for "hundred-meter resolution and minute-level updates" of short-term nowcast gust forecasts. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to overcome the shortcomings of the prior art and provide a method and device for objective forecasting of gusts at the 100-meter level, so as to solve the problem that the existing gust prediction methods cannot meet the rigid requirement of updating short-term and nowcast gusts at the 100-meter resolution.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for objective forecasting of 100-meter-level gusts, comprising:

[0006] Obtain weather element data;

[0007] The weather element data is quantitatively analyzed using multiple preset intervals to obtain a gust coefficient model; the gust coefficient model is used to represent the mapping relationship between average wind and gust under multiple preset intervals; the preset intervals include intervals of different altitudes, intervals of different wind speeds, and intervals of different wind directions;

[0008] The gust coefficient model is interpolated into a grid field with a resolution of 100 meters to obtain a grid field of gust coefficients; wherein, for each grid point, the gust coefficient corresponds to a different interval range.

[0009] The preset average wind is obtained based on the pre-acquired average style point deviation correction coefficient at a resolution of 100 meters.

[0010] Based on the preset average wind range and the gust coefficient grid field, obtain gust point prediction data with a resolution of 100 meters.

[0011] Furthermore, it also includes:

[0012] By coupling the gust coefficient grid field, the pre-acquired mean style point deviation correction coefficient at a resolution of 100 meters, and the pre-constructed gust machine learning model for high-altitude stations, gust prediction data for high-altitude stations are obtained.

[0013] Furthermore, it also includes:

[0014] The obtained gust point prediction data with a resolution of 100 meters and the gust mountain station prediction data were evaluated and tested according to preset standards.

[0015] Furthermore, the quantitative analysis of the weather element data using multiple preset intervals to obtain the gust coefficient model includes:

[0016] Acquire meteorological data for a preset time period from automatic weather stations in a designated area; the meteorological data includes average wind speed, wind direction, instantaneous wind speed, and station altitude information; wherein, the instantaneous wind speed is the maximum instantaneous wind speed within 1 hour, and the average wind speed is the average wind speed over a 2-minute period from the time the maximum instantaneous wind speed occurs;

[0017] The gust coefficients for different wind direction ranges, different wind speed ranges, and different altitude ranges are calculated based on the average wind speed and instantaneous wind speed.

[0018] All gust coefficients constitute the gust coefficient model.

[0019] Furthermore, methods for obtaining the average style point bias correction coefficient at a resolution of 100 meters include:

[0020] The station deviation coefficient is determined based on long-term historical data of predicted average wind speed and observed average wind speed.

[0021] The station deviation coefficients are interpolated onto a high-resolution grid field using the inverse distance interpolation method to obtain grid deviation correction coefficients with a resolution of hundreds of meters.

[0022] Furthermore, based on the preset average wind range and the gust coefficient grid field, gust point prediction data with a resolution of 100 meters is obtained, including:

[0023] Obtain the altitude, average wind speed, and average wind direction of the grid points corresponding to the preset average wind, and determine the interval to which the preset average wind belongs and the corresponding gust coefficient value.

[0024] Calculate gust prediction data based on the gust coefficient value and the preset average wind.

[0025] The gust prediction data refers to the gust prediction values ​​calculated in the gust prediction field.

[0026] Furthermore, the current gust observation data, mapping relationships, and pre-acquired grid bias correction coefficients at a resolution of 100 meters are coupled with a pre-constructed gust machine learning model for high-altitude stations to obtain gust prediction data for high-altitude stations, including:

[0027] Obtain sample feature data;

[0028] The sample data is divided into training dataset and test dataset according to the season and different forecast lead times;

[0029] A machine learning model for gusts was constructed using decision trees as the basis function and mean squared error as the objective function.

[0030] The training dataset is input into the gust machine learning model for training until the objective function converges, thus obtaining the gust machine learning model for high-altitude stations.

[0031] The test dataset was input into a machine learning model for high-altitude stations to conduct tests.

[0032] Furthermore, the evaluation and testing of the obtained gust point prediction data with a resolution of 100 meters and the gust mountain station prediction data according to preset standards includes:

[0033] Based on the gust analysis field and the predicted field, the mean absolute error and root mean square error are calculated with the observed values, and the mean absolute error and root mean square error are compared and evaluated with preset standards.

[0034] This application provides a 100-meter-level objective gust forecasting device, comprising:

[0035] The acquisition module is used to acquire weather element data;

[0036] The analysis module is used to perform quantitative analysis on the weather element data in multiple preset intervals to obtain a gust coefficient model; the gust coefficient model is used to represent the mapping relationship between average wind and gust under multiple preset intervals; the preset intervals include intervals of different altitudes, intervals of different wind speeds, and intervals of different wind directions;

[0037] An interpolation module is used to interpolate the gust coefficient model into a grid field with a resolution of hundreds of meters to obtain a gust coefficient grid field; wherein, for each grid point, the gust coefficient corresponds to a different interval range;

[0038] The pre-acquisition module is used to obtain the preset average wind based on the average style point deviation correction coefficient of the pre-acquisition 100-meter resolution.

[0039] The prediction data is used to obtain gust point prediction data with a resolution of hundreds of meters based on the range of the preset average wind and the grid field of the gust coefficient.

[0040] Furthermore, it also includes:

[0041] The coupling module is used to couple the gust coefficient grid field, the pre-acquired mean style point deviation correction coefficient with 100-meter resolution, and the pre-built gust machine learning model for high-altitude stations to obtain gust prediction data for high-altitude stations.

[0042] The beneficial effects that can be achieved by adopting the above technical solution in this invention include:

[0043] This invention provides a method and apparatus for objective gust forecasting at the 100-meter level. This application enables in-depth research on the evolution characteristics and variation patterns of near-surface gusts under different underlying surfaces, terrain heights, seasons, and weather systems in relevant regions, leading to a better understanding of the formation mechanisms related to non-convective instantaneous strong winds. On the other hand, by developing and implementing high-resolution gridded forecast products for gusts under complex terrain conditions in the Beijing-Tianjin-Hebei region, this application can further improve the level of refined gust forecasting under complex terrain conditions, thereby enhancing the safety operation and disaster prevention and mitigation capabilities of major events and cities in the capital and surrounding areas. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a schematic diagram illustrating the steps of the objective gust forecasting method at the 100-meter level of the present invention;

[0046] Figure 2 This is a schematic diagram of the pre-set solar gust analysis field of the present invention;

[0047] Figure 3 This is a schematic diagram of the gust forecast field for the preset date of this invention;

[0048] Figure 4 This is a schematic diagram of the objective gust forecasting device at the 100-meter level of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0050] The following describes a specific method and apparatus for objective forecasting 100-meter-level gusts, provided in an embodiment of this application, with reference to the accompanying drawings.

[0051] like Figure 1 As shown, the objective forecasting method for 100-meter-level gusts provided in this application embodiment includes:

[0052] S101, acquire weather element data;

[0053] The weather element data includes collected and organized automatic weather station observation data, altitude data, and data from Ruitu-Ruisi analysis and forecast field. It can be understood that these data include average wind and gusts, specifically the altitude of each station, the hourly maximum wind speed and wind direction of each station, the average wind speed and average wind direction corresponding to the time of occurrence of the hourly maximum wind at each station, and the high-resolution grid average wind field reanalyzed by Ruitu-Ruisi.

[0054] S102, quantitative analysis of the weather element data is performed using multiple preset intervals to obtain a gust coefficient model; the gust coefficient model is used to represent the mapping relationship between average wind and gust under multiple preset intervals; the preset intervals include intervals of different altitudes, intervals of different wind speeds, and intervals of different wind directions;

[0055] It is understood that in this application, the obtained weather element data are quantitatively analyzed for different altitude ranges, different wind speed ranges, and different wind direction ranges to obtain the corresponding gust coefficients. All the gust coefficients constitute a gust coefficient model. The gust coefficient model represents the mapping relationship between the average wind and the gust under multiple preset ranges. That is to say, when the average wind is known, the gust coefficient of the corresponding range can be determined for different ranges, and then the corresponding gust can be predicted based on the gust coefficient and the average wind.

[0056] It should be noted that the RMAPS-RISE system is a rapid-updating, seamless fusion and integrated forecasting system built upon the China Meteorological Administration's Beijing Rapid Update Cyclic Numerical Weather Prediction System (CMA Beijing Model, CMA-BJ) (formerly the North China Regional Rapid Update Cyclic Numerical Weather Prediction System (RMAPS)) and observational data from automatic weather stations and radar. It utilizes multi-source data fusion technology, bias correction technology, and high-resolution topographic downscaling technology. This system can provide high-resolution diagnostic analysis and forecasts of 0-24 hour precipitation, temperature, mean wind, and precipitation phase, covering the entire Beijing-Tianjin-Hebei region (spatial resolution 500m) and key areas (100km × 100km, covering the Zhangjiakou and Yanqing mountainous competition zones, spatial resolution 100m), with a temporal resolution of 10min. This application can obtain mean wind through the RMAPS-RISE system. It is understood that the RMAPS-RISE system is a mature technology in the existing field, and will not be elaborated upon further here.

[0057] S103, interpolate the gust coefficient model into a grid field with a resolution of 100 meters to obtain a grid field of gust coefficient; wherein, for each grid point, the gust coefficient corresponds to a different interval range;

[0058] The gust coefficient model is interpolated into a grid field with a resolution of 100 meters to obtain the gust coefficient grid field. For each grid point in the gust coefficient grid field, each gust coefficient corresponds to a different interval range. Therefore, after obtaining the interval where the average wind is located, the corresponding gust coefficient can be obtained. They are all in a one-to-one correspondence.

[0059] S104, the preset average wind is obtained based on the pre-acquired average style point deviation correction coefficient at a 100-meter resolution;

[0060] It is understood that the preset average wind in this application is based on the Ruito-Ruisi system. In order to minimize the impact of the Ruitoi system's average forecast accuracy on the gust coefficient model, the actual observation data of the average wind field of nearly 3,000 automatic weather stations at different altitudes in the Beijing-Tianjin-Hebei region of the Ruitoi system are combined with the high-precision average wind field forecast data of the Ruitoi system to calculate the average wind point deviation correction coefficient with a resolution of 100 meters. The average wind in the Ruitoi system is corrected by the average wind point deviation correction coefficient with a resolution of 100 meters to obtain a more accurate average wind, thereby making the accuracy of the gusts calculated subsequently higher.

[0061] S105, based on the preset average wind range and the gust coefficient grid field, obtain gust point prediction data with a resolution of 100 meters.

[0062] In this process, the gust coefficient in the gust coefficient grid field is determined by the interval where the preset average wind is located, and the gust is calculated by the preset average wind and the gust coefficient, which is the gust point prediction data.

[0063] The working principle of the 100-meter-level objective gust forecasting method is as follows: The technical solution provided in this application is mainly applied to the Beijing-Tianjin-Hebei region. It collects and organizes observational data from automatic weather stations, altitude data, and data from the Ruitu-Ruisi analysis and forecast field. Then, based on long-term series data, the Ruitu-Ruisi system is used to conduct in-depth research on the overall fine characteristics of near-surface gusts and gust coefficients under different underlying surfaces, topographic conditions, seasons, and weather systems across the entire Beijing-Tianjin-Hebei region. This research also examines the differences between typical regions and the evolution characteristics and similarities of near-surface gusts, average winds, and gust coefficients at different stations with similar altitudes in high mountains. This yields a gust coefficient model between the gust coefficient at Beijing-Tianjin-Hebei stations and meteorological and physical elements such as stable wind speed, wind direction, and topographic altitude. A preset average wind is obtained based on the pre-acquired 100-meter-level resolution average wind point deviation correction coefficient. Then, the corresponding gust coefficient in the gust coefficient grid field can be determined through the range of the preset average wind. Based on the gust coefficient and the preset average wind, the predicted gust can be obtained. This study integrates and couples the gust coefficient model of Beijing-Tianjin-Hebei region with gust observation data fusion correction technology and grid bias correction and other model post-processing correction technologies, and realizes 24-hour refined objective forecast of gusts covering the entire Beijing-Tianjin-Hebei region and key areas based on the RuiTu-Ruisi system.

[0064] In some embodiments, the objective forecasting method for 100-meter-level gusts provided in this application further includes:

[0065] By coupling the gust coefficient grid field, the pre-acquired mean style point deviation correction coefficient at a resolution of 100 meters, and the pre-constructed gust machine learning model for high-altitude stations, gust prediction data for high-altitude stations are obtained.

[0066] This application uses gust data interpolated from the Ruisi system to high-altitude stations and gust data observed by automatic weather stations to construct objective gust products for high-altitude stations using machine learning methods. It also evaluates and verifies the forecast performance of high spatiotemporal resolution gridded gust products and objective products for high-altitude stations, as well as the degree of influence of strong winds and model forecast accuracy. Ultimately, it can predict gust products for high-altitude stations.

[0067] In some embodiments, the step of quantitatively analyzing the weather element data using multiple preset intervals to obtain a gust coefficient model includes:

[0068] Acquire meteorological data for a preset time period from automatic weather stations in a designated area; the meteorological data includes average wind speed, wind direction, instantaneous wind speed, and station altitude information; wherein, the instantaneous wind speed is the maximum instantaneous wind speed within 1 hour, and the average wind speed is the average wind speed over a 2-minute period from the time the maximum instantaneous wind speed occurs;

[0069] The gust coefficients for different wind direction ranges, different wind speed ranges, and different altitude ranges are calculated based on the average wind speed and instantaneous wind speed.

[0070] All gust coefficients constitute the gust coefficient model.

[0071] Specifically, data on 2-minute average wind speed, wind direction, instantaneous wind speed, wind direction, and station altitude from national and regional automatic weather stations in the Beijing-Tianjin-Hebei region are collected at 10-minute intervals. Following the automatic weather station gust storage rules, the observed data is preprocessed into hourly intervals, storing the maximum instantaneous wind speed (maximum wind) WS from the previous hour to the current hour. X Wind direction WD X The 2-minute average wind speed WS corresponding to the time when the instantaneous wind maximum occurs. 2a Wind direction WD 2a Calculate the hourly gust coefficient GF for various stations in the Beijing-Tianjin-Hebei region. k The gust coefficient (maximum wind speed / average wind speed) is calculated as follows:

[0072]

[0073] Then, using statistical analysis methods, the gust coefficients of automatic weather stations at different altitudes in the Beijing-Tianjin-Hebei region were calculated for four different wind speed ranges (0-2; 3-5; 6; 7 and above) and eight different wind direction ranges (divided into eight directions: the first direction is northerly winds 337.5-360°, 0-22.5°; the second direction is northeasterly winds 22.5-67.5°, ..., increasing sequentially at 45° intervals). The gust coefficients were then comprehensively analyzed in relation to the climatological statistical characteristics and local characteristics of different terrain altitudes, wind directions, and wind speeds. All gust coefficients representing the climatological statistical characteristics and local characteristics of different terrain altitudes, wind directions, and wind speeds constituted a gust coefficient model.

[0074] In some embodiments, the method for obtaining the average style point bias correction coefficient at a resolution of 100 meters includes:

[0075] The station deviation coefficient is determined based on long-term historical data of predicted average wind speed and observed average wind speed.

[0076] The station deviation coefficients are interpolated onto a high-resolution grid field using the inverse distance interpolation method to obtain grid deviation correction coefficients with a resolution of hundreds of meters.

[0077] In some embodiments, obtaining gust point prediction data with a resolution of hundreds of meters based on the range of the preset average wind and the grid field of gust coefficients includes:

[0078] Obtain the altitude, average wind speed, and average wind direction of the grid points corresponding to the preset average wind, and determine the interval to which the preset average wind belongs and the corresponding gust coefficient value.

[0079] Calculate gust prediction data based on the gust coefficient value and the preset average wind.

[0080] The gust prediction data refers to the gust prediction values ​​calculated in the gust prediction field.

[0081] It should be noted that this application can improve the accuracy of prediction results by first establishing the gust analysis field and then establishing the gust prediction field. Specifically, the gust coefficient of each grid point (i,j) in the Ruito-Ruisi system is set to 1.8 as the gust coefficient background field. This is then interpolated onto the Ruito-Ruisi 100-meter resolution grid field using the bilinear distance inverse interpolation method. The statistically obtained gust coefficient is taken as the true value. The error between the gust coefficient background field and the true value is first determined by the difference between the observed value and the background field value of neighboring grid points, and then the error on other grid points is determined by the bilinear distance inverse weighting. For each grid point (i,j) of the Ruito-Ruisi 100-meter resolution, the distance to the kth automatic weather station is r. ijk Using the inverse bilinear distance ratio The interpolation method correlates the gust coefficient GF of the k-th automatic weather station in the Beijing-Tianjin-Hebei region with meteorological elements such as stable wind speed, wind direction, and terrain elevation. k Interpolating the gust coefficient to each grid point (i,j) of the Ruito-Ruisi system and subtracting it from the background field yields the gust coefficient difference field ΔGF(i,j) for each grid point in the Ruito-Ruisi system for the Beijing-Tianjin-Hebei region. Adding the background field and the difference field gives the gust coefficient grid field GF(i,j), calculated as follows.

[0082]

[0083] GF(i,j)=1.8+ΔGF(i,j) (3)

[0084] Where n is the total number of the nearest automatic weather stations used in the interpolation, and in the technical solution provided in this application, n is taken as 8. For each grid point, the gust coefficient corresponds to 32 different interval ranges (e.g., for the Beijing Observatory station, cases 1-8 represent the gust coefficient values ​​corresponding to 8 different wind directions when the wind speed is below level 3, cases 9-16 represent the gust coefficient values ​​corresponding to 8 different wind directions when the wind speed is level 3-5, ..., and so on). During real-time operation, the interval to which the gust coefficient belongs and its corresponding value are determined based on the average wind speed and wind direction of the analysis field or forecast field corresponding to the (i,j) grid point in the RuiTu-Ruisi system read in real time (e.g., for a certain grid point (3,10), its average wind speed is 3m / s and the wind direction is 220°).

[0085] When constructing the Ruitu-Ruisi gust analysis field at time t0, in order to better integrate and absorb gust observation data, the background field is based on the average maximum wind value of the Ruitu-Ruisi system at 10-minute intervals within each hour. (For example, when calculating the gust analysis field reported from 00:30, the maximum value of the mean wind speed of the Ruito-Ruisi system at four time points (00:00, 00:10, 00:20, and 00:30) is selected as the background field. This is coupled and integrated with the gridded field GF(i,j) of the gust coefficient in the Beijing-Tianjin-Hebei region as the initial guess field. The initial guess field is then combined with the gust automatic weather station observation data X. k OBS The error is first determined by the difference between the initial guess value and the ground observation value of the neighboring grid points. Then, the error ΔX(i,j) on other grid points is determined by the inverse distance weighting. The initial guess field and the gust wind field difference field are added together to obtain the gust analysis field. The calculation is performed using the following method.

[0086]

[0087]

[0088] In the formula, Here, ANA represents the gust analysis value at time t0, and ANA indicates the analysis field. Let GF(i,j) be the average maximum wind speed of the Ruito-Ruisi system at 10-minute intervals within one hour at grid point (i,j), GF(i,j) be the gust coefficient at each grid point (i,j) for 32 different intervals, and ΔX(i,j) be the gust difference at grid point (i,j). k OBS The gust wind observation value is from the automatic weather station at the k-th station. The values ​​represent the gust values ​​at neighboring grid points. Finally, the resulting gust analysis field is as follows: Figure 2 As shown.

[0089] When constructing the gust forecast field, the first step is to correct the t value after grid bias. i Forecast time of the RuiTu-Ruisi average wind forecast field As the initial field of prediction, FORC represents the forecast field. By reading the average wind speed and direction at each grid point (i,j), the interval to which the gust coefficient belongs and its corresponding value GF(i,j) are determined, and then multiplied by t. i The forecast time is t. The mean wind forecast field of RuiTu-Ruisi is used to obtain the gust forecast field. i Gust forecast field The calculation method is as follows, and finally, the gust forecast field is obtained as follows: Figure 3 As shown.

[0090]

[0091] To minimize the impact of the average forecast accuracy of the Ruisi system on the gust coefficient model, the average wind field observation data from nearly 3,000 automatic weather stations at different altitudes in the Beijing-Tianjin-Hebei region were combined with the high-precision average wind field forecast data from the Ruisi system. Statistical bias correction methods were used to obtain the data for each station and each forecast lead time (t) in the Beijing-Tianjin-Hebei region under complex terrain conditions. i The average wind speed predicted by Ruisi for different wind force levels (0-2; 3-5; 6; 7 and above). Compared with the observed average wind speed The ratio of the two values ​​is defined as the station deviation coefficient S. k Then S k Interpolation by inverse distance (1 / r) 2 The method interpolates the data onto the high-resolution grid field of the Ruisi database to obtain the grid bias correction coefficient C. ij In this application, when obtaining the gust forecast field, the grid bias correction coefficient C is used. ij The average wind speed of the Ruito-Ruisi system is corrected to make the average wind speed more accurate. The specific calculation method is as follows:

[0092]

[0093]

[0094] In some embodiments, the coupling of current gust observation data, mapping relationships, and pre-acquired grid bias correction coefficients at a resolution of 100 meters with a pre-constructed gust machine learning model for high-altitude stations to obtain gust prediction data for high-altitude stations includes:

[0095] Obtain sample feature data;

[0096] The sample data is divided into training dataset and test dataset according to the season and different forecast lead times;

[0097] A machine learning model for gusts was constructed using decision trees as the basis function and mean squared error as the objective function.

[0098] The training dataset is input into the gust machine learning model for training until the objective function converges, thus obtaining the gust machine learning model for high-altitude stations.

[0099] The test dataset was input into a machine learning model for high-altitude stations to conduct tests.

[0100] Specifically, for high-altitude stations in certain regions, this application, based on grid bias correction, further develops a high-altitude station gust wind field bias correction technique using the XGBoost machine learning method. The machine learning model sample data uses long-term series Ruitu-Ruisi system interpolated to gust wind data (UGUST, VGUST, WSGUST, WDGUST) for high-altitude stations in the corresponding region (e.g., Zhangjiakou) and observation data for the corresponding time periods at the corresponding stations. The machine learning model sample features include 26 features, such as: the difference d between the forecast and observation at different start times of the previous day in the Ruisi system. n t (where n is the forecast lead time, t is the current forecast start time, and d is the forecast duration.) n t =O n t -F n t The forecast value F for the next time of reporting. n t+1 The altitude H of each station stid The truth label is the observation O at the next time step. t+1 .

[0101] Table 1. Model Sample Characteristics

[0102] <![CDATA[d n t-23 ]]> <![CDATA[d n t-22 ]]> <![CDATA[d n t-21 ]]> …… <![CDATA[d n t-1 ]]> <![CDATA[d n t ]]> <![CDATA[F n t+1 ]]> <![CDATA[H stid ]]> <![CDATA[O t+1 ]]>

[0103] When building a machine learning model, training and testing datasets are allocated according to season and different forecast lead times. Decision trees are used as base functions, and the mean squared error (MSE) is used as the objective function. The training process of the XGBoost model involves finding the optimal parameter set by minimizing the objective function. Its objective function is shown below:

[0104]

[0105] XGBoost employs additive training, meaning that the model's convergence objective is not to directly optimize the entire objective function, but rather to optimize the objective function distributedly, as shown in the following equation:

[0106]

[0107]

[0108]

[0109] Substituting (11) into (10), we can obtain the objective function of the model when training the t-th decision tree as follows:

[0110]

[0111] Where n represents the total number of samples, and t represents the t-th decision tree. It is a loss function. This is the regularization term. Substituting the MSE loss function into the above equation, we finally obtain the final form of the objective function when optimizing the t-th decision tree, which is:

[0112]

[0113] In some embodiments, it also includes:

[0114] The obtained gust point prediction data with a resolution of 100 meters and the gust mountain station prediction data were evaluated and tested according to preset standards.

[0115] As a preferred embodiment, the evaluation and testing of the obtained gust point prediction data with a resolution of 100 meters and the gust mountain station prediction data according to preset standards includes:

[0116] Based on the gust analysis field and the predicted field, the mean absolute error and root mean square error are calculated with the observed values, and the mean absolute error and root mean square error are compared and evaluated with preset standards.

[0117] Specifically, this application uses conventional statistical measures such as mean absolute error (MAE) and root mean square error (RMSE) to objectively verify and evaluate gust forecasts. In addition, to assess the impact of strong winds on gust products, wind speed and direction forecast scores are calculated for different wind speed level ranges. To assess the impact of complex terrain on gust forecast data, mean absolute error and root mean square error are calculated for high-altitude stations and representative plain stations at different altitudes. To assess the impact of model forecast accuracy on gust products, mean absolute error and root mean square error are calculated for products coupled with the Ruisi mean wind field and gust coefficient model before and after correction. The verification standard refers to the China Meteorological Administration's wind forecast verification standard "QXT 229—2014 Wind Forecast Verification Method".

[0118] like Figure 4 As shown in the figure, this application provides a 100-meter-level objective gust forecasting device, comprising:

[0119] Module 201 is used to acquire weather element data;

[0120] Analysis module 202 is used to perform quantitative analysis on the weather element data in multiple preset intervals to obtain a gust coefficient model; the gust coefficient model is used to represent the mapping relationship between average wind and gust under multiple preset intervals; the preset intervals include intervals of different altitudes, intervals of different wind speeds, and intervals of different wind directions;

[0121] Interpolation module 203 is used to interpolate the gust coefficient model into a grid field with a resolution of hundreds of meters to obtain a gust coefficient grid field; wherein, for each grid point, the gust coefficient corresponds to a different interval range;

[0122] The pre-acquisition module 204 is used to obtain the preset average wind based on the average style point deviation correction coefficient of the pre-acquisition 100-meter resolution.

[0123] The prediction module 205 is used to obtain gust point prediction data with a resolution of 100 meters based on the range of the preset average wind and the grid field of the gust coefficient.

[0124] The objective gust forecasting device with a 100-meter level provided in this application embodiment also includes:

[0125] The coupling module is used to couple the gust coefficient grid field, the pre-acquired mean style point deviation correction coefficient with 100-meter resolution, and the pre-built gust machine learning model for high-altitude stations to obtain gust prediction data for high-altitude stations.

[0126] The working principle of the 100-meter-level objective gust forecasting device provided in this application is as follows: the acquisition module 201 acquires weather element data; the analysis module 202 performs quantitative analysis on the weather element data using multiple preset intervals to obtain a gust coefficient model; the gust coefficient model is used to represent the mapping relationship between average wind and gust under multiple preset intervals; the preset intervals include intervals of different altitudes, different wind speeds, and different wind directions; the interpolation module 203 interpolates the gust coefficient model into a 100-meter-level resolution grid field to obtain a gust coefficient grid field; wherein, for each grid point, the gust coefficient corresponds to a different interval range; the pre-acquisition module 204 obtains a preset average wind based on the pre-acquisition 100-meter-level resolution average wind point deviation correction coefficient; and the prediction module 205 obtains 100-meter-level resolution gust point prediction data based on the interval range of the preset average wind and the gust coefficient grid field.

[0127] In summary, this invention provides a method and apparatus for objective gust forecasting at the 100-meter level. This invention constructs a gust coefficient model and combines it with a 100-meter resolution mean wind point deviation correction coefficient to obtain the corrected mean wind from the Ruisi system, thus obtaining the predicted gust. This application can improve the level of refined gust forecasting under complex terrain conditions, and enhance the safety operation and disaster prevention and mitigation capabilities of major events and cities in the capital and surrounding areas.

[0128] It is understood that the method embodiments provided above correspond to the device embodiments described above, and the specific details can be referred to each other, which will not be repeated here.

[0129] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0130] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0131] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction methods implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0132] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0133] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for objectively forecasting gusts at the 100-meter level, characterized in that, include: Obtain weather element data; A gust coefficient model is obtained by quantitatively analyzing the weather element data across multiple preset intervals, including: acquiring meteorological data from automatic weather stations in a designated area over a preset time interval; the meteorological data includes average wind speed, wind direction, instantaneous wind speed, and station altitude information; wherein the instantaneous wind speed is the maximum instantaneous wind speed within one hour, and the average wind speed is the 2-minute average wind speed at the time the maximum instantaneous wind speed occurs; calculating the gust coefficient for different wind direction intervals, different wind speed intervals, and different altitude intervals based on the average wind speed and instantaneous wind speed; all gust coefficients constitute the gust coefficient model; wherein the gust coefficient model is used to represent the mapping relationship between average wind and gusts across multiple preset intervals; the preset intervals include intervals for different altitudes, different wind speeds, and different wind directions; The gust coefficient model is interpolated into a grid field with a resolution of 100 meters to obtain a grid field of gust coefficients; wherein, for each grid point, the gust coefficient corresponds to a different interval range. The preset average wind speed is obtained based on the pre-acquired average wind point deviation correction coefficient at a resolution of 100 meters; wherein, the method for obtaining the average wind point deviation correction coefficient at a resolution of 100 meters includes: determining the station deviation coefficient based on long-term historical data of predicted average wind speed and observed average wind speed; interpolating the station deviation coefficient onto the high-resolution grid field using the inverse distance interpolation method to obtain the grid deviation correction coefficient at a resolution of 100 meters. Based on the preset average wind range and the gust coefficient grid field, obtain gust point prediction data with a resolution of 100 meters.

2. The method according to claim 1, characterized in that, Also includes: By coupling the gust coefficient grid field, the pre-acquired mean style point deviation correction coefficient at a resolution of 100 meters, and the pre-constructed gust machine learning model for high-altitude stations, gust prediction data for high-altitude stations are obtained.

3. The method according to claim 1 or 2, characterized in that, Also includes: The obtained gust point prediction data with a resolution of 100 meters and the gust mountain station prediction data were evaluated and tested according to preset standards.

4. The method according to claim 1, characterized in that, Based on the preset average wind range and the grid field of gust coefficients, obtain gust point prediction data with a resolution of 100 meters, including: Obtain the altitude, average wind speed, and average wind direction of the grid points corresponding to the preset average wind, and determine the interval to which the preset average wind belongs and the corresponding gust coefficient value. Calculate gust prediction data based on the gust coefficient value and the preset average wind. The gust prediction data refers to the gust prediction values ​​calculated in the gust prediction field.

5. The method according to claim 2, characterized in that, The gust coefficient grid field, the pre-acquired mean style point deviation correction coefficient at a 100-meter resolution, and the pre-constructed gust machine learning model for high-altitude stations are coupled to obtain gust prediction data for high-altitude stations, including: Obtain sample feature data; The sample feature data are divided into training datasets and test datasets according to the season and different forecast lead times; A machine learning model for gusts was constructed using decision trees as the basis function and mean squared error as the objective function. The training dataset is input into the gust machine learning model for training until the objective function converges, thus obtaining the gust machine learning model for high-altitude stations. The test dataset was input into a machine learning model for high-altitude stations to conduct tests.

6. The method according to claim 3, characterized in that, The evaluation and testing of the obtained gust point prediction data with a resolution of 100 meters and the gust mountain station prediction data are performed according to preset standards, including: Based on the gust analysis field and the predicted field, the mean absolute error and root mean square error are calculated with the observed values, and the mean absolute error and root mean square error are compared and evaluated with preset standards.

7. A 100-meter-level objective gust forecasting device, characterized in that, include: The acquisition module is used to acquire weather element data; The analysis module is used to quantitatively analyze the weather element data across multiple preset intervals to obtain a gust coefficient model. This gust coefficient model represents the mapping relationship between average wind and gusts across the multiple preset intervals. The preset intervals include intervals for different altitudes, different wind speeds, and different wind directions. Specifically, it is used to acquire meteorological data from automatic weather stations in a designated area over preset time intervals. The meteorological data includes average wind speed, wind direction, instantaneous wind speed, and station altitude information. The instantaneous wind speed is the maximum instantaneous wind speed within one hour, and the average wind speed is the 2-minute average wind speed at the time the maximum instantaneous wind speed occurs. Based on the average wind speed and instantaneous wind speed, the gust coefficients for different wind direction intervals, different wind speed intervals, and different altitude intervals are calculated. All gust coefficients constitute the gust coefficient model. An interpolation module is used to interpolate the gust coefficient model into a grid field with a resolution of hundreds of meters to obtain a gust coefficient grid field; wherein, for each grid point, the gust coefficient corresponds to a different interval range; The pre-acquisition module is used to obtain a preset average wind based on the pre-acquisition of the average wind point deviation correction coefficient at a resolution of 100 meters; wherein, the method for obtaining the average wind point deviation correction coefficient at a resolution of 100 meters includes: determining the station deviation coefficient based on long-term historical data of predicted average wind speed and observed average wind speed; interpolating the station deviation coefficient onto a high-resolution grid field using the inverse distance interpolation method to obtain the grid deviation correction coefficient at a resolution of 100 meters. The prediction module is used to obtain gust point prediction data with a resolution of 100 meters based on the range of the preset average wind and the grid field of the gust coefficient.

8. The apparatus according to claim 7, further comprising: The coupling module is used to couple the gust coefficient grid field, the pre-acquired mean style point deviation correction coefficient with 100-meter resolution, and the pre-built gust machine learning model for high-altitude stations to obtain gust prediction data for high-altitude stations.

Images