A method, medium, and program product for representative evaluation of regional wind speed weather reference stations based on a correlation function field

Abstract

The application discloses a kind of representative evaluation methods of regional wind speed meteorological reference station based on correlation function field, medium and program product, comprising: collecting target area high-density meteorological observation site multi-year daily wind speed data and the environmental information of observation point;Select the independent sample of each same time period daily wind speed data in region;Establish the correlation function field between two sites, fit the relationship model of correlation function and distance;Estimate the standardization observation random mean square error;Calculate the relative standard deviation generated by reference station wind speed observation data instead of target point wind speed true value data;Evaluate the excellent grade of the spatial distance range that reference station wind speed can represent.The application can realize that reference station has good representative evaluation in what distance range, and has important practical value for selecting reliable meteorological reference station in climate feasibility demonstration selection such as wind power development site selection, urban and rural construction planning, evaluation extreme climate risk and the accuracy of engineering meteorological parameter calculation.

Description

Technical Field

[0001] This invention belongs to the field of meteorological applications and assessments, and involves methods, media, and program products for assessing the representativeness of regional meteorological reference stations. Specifically, it involves calculating the relative standard error that may arise when using wind speed observation data of a certain station as a substitute for wind speed data of a meteorological reference station based on the wind speed correlation function field between various stations in a region. This determines the representativeness level of different distance ranges that the wind speed meteorological reference station can represent, thereby improving the accuracy of selecting and using meteorological reference stations in climate feasibility studies and climate risk assessments for engineering projects. Background Technology

[0002] Meteorological reference stations, as crucial sources of regional meteorological data, play an irreplaceable role in numerous industries, including wind energy resource development, engineering climate assessment, urban planning, and disaster risk assessment. For example, in wind farm site selection, long-term monitoring data on wind speed and direction from reference stations is used to analyze the spatiotemporal distribution characteristics of wind energy resources and determine the development potential of the region's wind energy. In transmission line design, data such as maximum wind speed and icing thickness from meteorological reference stations are vital. Based on extreme meteorological parameters provided by the stations, the strength of transmission line towers and the selection of conductors are rationally designed to ensure safe operation under severe weather conditions such as strong winds and icing. In urban planning, data on wind speed, precipitation, and temperature from meteorological reference stations can be used for ventilation corridor design, storm precipitation formula development, and urban functional layout. In agricultural planting planning, meteorological data from meteorological reference stations can analyze the climate adaptability of different crops, guiding adjustments to agricultural planting structures and optimizing regional layouts.

[0003] The representativeness assessment of meteorological reference stations is a crucial step in ensuring that meteorological data (such as wind speed, precipitation, and temperature) can effectively support applications such as engineering design, climate analysis, and disaster assessment. The core issue is determining whether the meteorological characteristics of the reference station are consistent with or highly similar to those of the target area. Existing meteorological reference station selection and assessment techniques typically fall into two categories: qualitative assessment, which relies primarily on factors such as spatial distance, climate zoning, and underlying surface characteristics. However, this approach lacks quantitative standards and depends heavily on subjective judgment based on expert experience, making it difficult to guarantee the objectivity and reliability of the assessment results. Quantitative analysis involves establishing a dedicated meteorological observation station in the target area or finding a short-sequence meteorological station nearby and conducting correlation analysis with the reference station to determine its suitability. However, this method requires at least one year of observational data from the target area and does not consider potential errors from the reference station. Therefore, currently used meteorological reference station selection and assessment methods have certain shortcomings or deficiencies.

[0004] As major project construction and development plans place increasingly higher demands on the precision required to address extreme weather events, higher requirements are being placed on the selection and evaluation of meteorological reference stations to determine their representativeness of the meteorological conditions in the project area. Furthermore, with the accelerated pace of meteorological modernization, various regions have constructed high-density provincial automatic weather stations to meet disaster prevention and mitigation needs. This lays the foundation for establishing correlation function fields of meteorological elements in the target area as a function of distance. Therefore, researching methods to quantitatively calculate the relative standard error of wind speed data generated by meteorological reference stations replacing target points based on regional correlation function fields, thereby determining the error of target points represented by wind speed meteorological reference stations at different distances and their representativeness level, and effectively improving the accuracy of meteorological reference station selection and evaluation, is a key technical problem that urgently needs to be solved. Summary of the Invention

[0005] (a) Purpose of the invention

[0006] Existing methods for selecting and evaluating meteorological reference stations, whether qualitative assessments based on spatial distance, the same climate zone, and underlying surface characteristics, or correlation analysis between meteorological observation stations within the target area and the proposed reference stations, fail to consider the data error issues that may arise from substituting reference station data for target point data. Clearly, both methods have technical shortcomings. To address at least one of these technical problems, this invention aims to provide a representative evaluation method, medium, and program product for regional wind speed meteorological reference stations based on a correlation function field. This method involves selecting independent samples of element data from each station within the region for the same time period. A correlation function field is established between each pair of stations within the region. Then, a polynomial fitting model of the relationship between the correlation function and distance is used to estimate the standardized observation random mean square error (MSE). or v The relative standard deviation (RSD) resulting from the use of meteorological reference station data instead of target point data. E r Then, based on the relative standard deviation ( E r ) is not greater than a certain standardized observation random mean square error ( or v The threshold is used to classify the representativeness and quality standards of wind speed data from reference stations, thereby achieving a representative assessment of the spatial distance range of wind speed meteorological reference stations. This provides support for selecting reliable meteorological reference stations to calculate engineering design parameters and assess the impact of meteorological disasters on projects.

[0007] (II) Technical Solution

[0008] To achieve the objective of this invention and solve its technical problems, the present invention adopts the following technical solution:

[0009] The first objective of this invention is to provide a method for evaluating the representativeness of regional wind speed meteorological reference stations based on a correlation function field. The representativeness of the wind speed meteorological reference station refers to the degree to which the wind speed observations of the reference station can reflect the true wind speed value at a target point with a certain spatial distance. The innovation of this method lies in its ability to quantitatively calculate the relative standard error (RSI) of the wind speed data generated by the meteorological reference station replacing the target point based on the regional correlation function field. E r ), and based on the relative standard error ( E r Not greater than the standardized observation random mean square error or v Determining the representativeness and quality level of wind speed meteorological reference stations can effectively improve the accuracy of selecting and evaluating meteorological reference stations at different distances from the target point, providing support for selecting reliable meteorological reference stations for calculating engineering design parameters and assessing the impact of meteorological disasters on projects. The method, when implemented, includes at least the following steps:

[0010] S100. Data Collection and Preprocessing: Collect geographical environmental information and multi-year daily wind speed observation data from high-density meteorological observation stations within the target assessment area that belong to the same climate zone and have wind speed observation capabilities. Perform quality control and preprocessing on the collected wind speed data to form a wind speed element sample database.

[0011] S200. Independent Sample Selection: Based on the preprocessed wind speed data, for the daily wind speed observation data of each meteorological observation station in the target assessment area for each year, independent samples are selected at set time intervals according to the same time series within each preset assessment period to ensure that the number of independent samples of each station in each assessment period is not less than the preset number, forming a wind speed element sample set with statistical independence.

[0012] S300. Construction of Regional Correlation Function Field: Based on the constructed wind speed element sample set, the distances between all stations and the wind speed observation correlation functions are statistically analyzed to obtain the relationship field between the distances between stations and the corresponding correlation functions; the relationship field is processed by moving average according to the distance to obtain the average correlation function value corresponding to each distance segment, and a regression model of the wind speed observation correlation function r′(d) with respect to the distance d is established accordingly;

[0013] S400. Estimation of the random mean square error of observations: Under the condition that the target assessment area satisfies the homogeneity and isotropy of the same climate zone, establish the analytical relationship between the wind speed observation correlation function r′(d) and the theoretical true value correlation function r(d), and based on this, calculate the standardized random mean square error of observations at d=0 using the intercept a0 of the regression model. or v It is used to measure the relative level of random error in regional wind speed observations;

[0014] S500. Calculation of Reference Station Substitution Error: Based on the regression model of wind speed observation correlation function with respect to distance and the standardized random mean square error of observations, the relative standard error generated by the replacement of the true wind speed value of the target point with the wind speed observation data of candidate reference stations in the evaluation area is calculated. E r ;

[0015] S600. Representativeness Level Classification: Based on Standardized Observational Random Mean Square Error or v and the relative standard error of the reference station E r Based on the threshold range, the representativeness of the reference stations is divided into several levels from excellent to poor, and each level corresponds to different engineering application suggestions;

[0016] S700. Spatial Distance Inversion and Representativeness Assessment: The spatial distance range corresponding to each representativeness level threshold is calculated through inversion. Based on this, the representativeness and quality of the spatial range of wind speed meteorological reference stations can be determined and assessed, providing a basis for the selection of reference stations for engineering applications with different accuracy requirements.

[0017] The second objective of this invention is to provide a computer program product, including computer instructions, for executing the above-mentioned regional wind speed meteorological reference station representativeness assessment method based on correlation function field.

[0018] The third objective of this invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for evaluating the representativeness of regional wind speed meteorological reference stations based on a correlation function field.

[0019] (III) Technical Effects

[0020] Compared with the prior art, the regional wind speed meteorological reference station representativeness assessment method, medium, and program product based on the correlation function field of the present invention have the following beneficial and significant technical effects:

[0021] Existing meteorological reference station selection and evaluation techniques typically employ two methods: one involves qualitative criteria followed by expert judgment; the other utilizes quantitative correlation analysis. Since good correlation only reflects "trend synchronicity" and does not necessarily imply small deviation, a more important indicator of whether a meteorological reference station represents a target point is the magnitude of the deviation between the two. Conventional meteorological reference station selection and evaluation methods suffer from a major deficiency: they cannot determine or lack suitable evaluation deviation threshold standards. This invention constructs a correlation function field for regional wind speed and uses the relationship between the correlation function and deviation to calculate the standardized observational random mean square error (RMSE) of the evaluation area. Then, based on the principle that the data deviation caused by distance factors for the reference station replacing the target point should not exceed a certain standardized observational RMSE threshold, a standard for classifying the representativeness level of reference stations is established. This answers the question of which time periods and spatial distance ranges provide good representativeness for meteorological reference stations. This has significant economic implications for fully utilizing reference station data, accurately calculating engineering meteorological parameters, and reducing engineering costs. Furthermore, the representativeness results of the reference stations evaluated using this method can quantify the representativeness error, making the selection and evaluation of meteorological reference stations more objective, and possessing significant scientific and application value. Attached Figure Description

[0022] Figure 1 The diagram shows the implementation process of the regional wind speed meteorological reference station representative evaluation method based on the correlation function field of the present invention.

[0023] Figure 2 The diagram shows the relationship between the correlation function of the average winter wind speed in southern Hebei Province calculated using the method of this invention and distance.

[0024] Figure 3 The diagram shows the relationship between the correlation function of the average spring wind speed in southern Hebei and distance, calculated using the method of this invention.

[0025] Figure 4 The diagram shows the relationship between the correlation function of the average summer wind speed in southern Hebei calculated using the method of this invention and distance.

[0026] Figure 5 The diagram shows the relationship between the correlation function of the average autumn wind speed in southern Hebei and distance, calculated using the method of this invention.

[0027] Figure 6 The diagram shows the variation of the relative standard error with distance in the four seasons of southern Hebei Province, calculated using the method of this invention. Detailed Implementation

[0028] The purpose of this invention is to provide a method, medium, and program product for evaluating the representativeness of regional wind speed meteorological reference stations based on a correlation function field. This method is used to quantitatively determine the representativeness level and spatial applicability of wind speed observation data from reference stations at different spatial distances to the true wind speed at a target point within a target area. The technical solutions in the embodiments of this invention will be described in more detail below with reference to the accompanying drawings. The described embodiments are only some, not all, of the embodiments of this invention.

[0029] Example 1: Representative Evaluation Method for Wind Speed ​​Meteorological Reference Stations

[0030] like Figure 1 As shown in the embodiment of the present invention, the representativeness assessment method for regional wind speed meteorological reference stations based on correlation function fields mainly includes the following steps during implementation:

[0031] S100. Data Collection and Preprocessing:

[0032] Geographical environmental information and multi-year daily wind speed observation data of high-density meteorological observation stations belonging to the same climate zone within the target assessment area are collected. The collected wind speed data are then subjected to quality control and preprocessing to form a wind speed element sample database.

[0033] Specifically, this step, in collecting and preprocessing the geographic environment information and wind speed observation data of the stations within the target assessment area, includes at least the following sub-steps:

[0034] Collect high-density meteorological observation stations within the target assessment area that belong to the same climate zone and have wind speed observation capabilities. High-density meteorological observation stations refer to national meteorological stations, provincial meteorological stations, etc. within the area, and their distribution density should be less than 25km apart.

[0035] Collect geographical environment information and daily wind speed observation data of each meteorological station. The geographical environment information of the station includes the longitude, latitude, altitude and underlying surface environment information of the station, and collect daily wind speed observation data of each station for 6 consecutive years or more.

[0036] The raw wind speed observation data collected from each station were preprocessed. The climate threshold check method was used to remove false values ​​in the raw data that exceeded the climate threshold. The spatiotemporal consistency check was used to identify and remove dead or erroneous values.

[0037] Spatial correlation statistical tests were performed on the preprocessed wind speed data of each station. The correlation coefficient between each station and the wind speed observation data of all stations within a 25km radius was calculated. Stations whose correlation coefficients with more than half of the stations within a 25km radius failed the 0.1 confidence test were removed.

[0038] A reliable wind speed element database will be established based on the stations that have completed quality control and correlation testing and their wind speed observation data. This database will include at least the station's geographical environment information and daily wind speed observation data.

[0039] S200. Independent Sample Selection:

[0040] Based on the preprocessed wind speed data, for the daily wind speed observation data of each meteorological observation station in the target assessment area for each year, independent samples are selected at set time intervals according to the same time series within each preset assessment period to ensure that the number of independent samples of each station in each assessment period is not less than the preset number, thus forming a wind speed element sample set with statistical independence.

[0041] In practical implementation, based on step S100, a reliable wind speed element database is established, and 50 or more independent samples of wind speed data from various stations within the region for the same time period are selected. To ensure the independence of the selected meteorological element samples and avoid the influence of weather system cycles, it is advisable to select one sample every 3 days or more from daily wind speed data with complete data for more than 6 years, and ensure that the number of samples for the evaluation period (month, quarter, year) is more than 50.

[0042] More specifically, this step, when selecting independent samples from each site within the target assessment area for a pre-defined assessment time period, includes at least the following sub-steps:

[0043] Based on the temporal variation characteristics of meteorological elements and the needs of engineering applications, the preset assessment period is divided into three types: monthly, seasonal, and annual. The seasonal boundaries and cross-year connection rules are clearly defined to ensure the comparability and summarization of sample sets under different time scales.

[0044] To ensure the independence of the selected meteorological element samples and avoid the influence of weather system cycles, a fixed starting point and fixed step size sampling strategy is adopted. The minimum sampling interval is limited to reduce the influence of weather-scale autocorrelation (e.g., a sample is selected every 3 days or more). The samples are repeatedly drawn in the same phase (date) on the multi-year series to ensure interannual comparability and statistical independence. The number of independent samples for each station in each evaluation period (month, quarter, year) is not less than the preset threshold (e.g., more than 50 groups, and at least 6 complete years).

[0045] Based on the independent samples obtained from each site and each evaluation period, a statistically independent wind speed element sample set is constructed, which includes at least the site code, sample time identifier, and wind speed observation value.

[0046] As a preferred approach, the three pre-defined evaluation periods—monthly, quarterly, and annual—are used during sampling:

[0047] For monthly time-period evaluation sampling, a strategy of selecting a sample group every 3 days is adopted (e.g., the 1st, 4th, 7th... 28th of each month can be selected). If there is missing data, the next day is selected to ensure that no less than 9 independent samples are obtained each month. The sampling is repeated in the same phase in the multi-year series (e.g., 54 samples are selected in 6 years) until the minimum number of samples required to meet the statistical analysis is obtained.

[0048] For seasonal sampling, the year is divided into four seasonal periods: winter (January, February, and December), spring (March, April, and May), summer (June, July, and August), and autumn (September, October, and November). A strategy of selecting a sample every 10 days is adopted, for example, the first day of each ten-day period of each month. If there is a missing data, the next day is selected to ensure that no less than 9 independent samples are obtained for each season. The sampling is repeated in the same phase in the multi-year series (for example, 54 samples are taken in 6 years) until the minimum number of samples required for statistical analysis is obtained.

[0049] For annual time-period evaluation sampling, a strategy of selecting a group of samples on the same date each month is adopted to ensure that no less than 12 independent samples are obtained each year. The samples are repeatedly drawn in the same phase over a multi-year series (e.g., 72 samples in 6 years) until the minimum number of samples required to satisfy the statistical analysis is obtained.

[0050] S300. Construction of the region-related function field:

[0051] Based on the constructed wind speed element sample set, the distances between all stations and the wind speed observation correlation functions are statistically analyzed to obtain the relationship field between the distances between stations and the corresponding correlation functions. The relationship field is then processed by moving average according to the distance to obtain the average correlation function value corresponding to each distance segment. Based on this, a regression model of the wind speed observation correlation function r′(d) with respect to distance d is established.

[0052] Specifically, when constructing the relationship field of distances and corresponding correlation functions between stations within the target evaluation area, at least the following sub-steps are included:

[0053] Based on the geographical environment information of each relevant meteorological observation station collected in step S100, the spatial distance between each pair of all stations is calculated to obtain the distance data of N=k×(k-1) / 2 pairs of stations, where k is the number of relevant stations in the region.

[0054] Based on the data sample selected for the evaluation period in step S200, the Pearson correlation coefficient formula was used for calculation. (Hereinafter referred to as Formula 1.1) Calculate the correlation function of wind speed observations between any two stations, where A and B represent any two stations within the region. r v(A,B) is the wind speed observation correlation function between stations A and B, used to characterize the degree of correlation between the wind speed observation data of the two stations, and d is the distance between stations A and B. and D represents the time series average of the wind speed deviations at stations A and B, respectively. v (A) and D v (B) represents the wind speed variances at stations A and B, respectively, with "—" indicating the average of the time series.

[0055] The calculated spatial distances between any two stations within the region and their corresponding wind speed observation correlation functions are paired and combined to form a field of relationships between station distances and their corresponding correlation functions.

[0056] Furthermore, when constructing a regression model of the wind speed observation correlation function r′(d) with respect to the inter-site distance d, at least the following sub-steps are included:

[0057] Based on N sets of data in the relationship field consisting of the spatial distance between each pair of stations and the corresponding wind speed observation correlation function, a 3km interval moving average is performed to reduce the fluctuation of the correlation function caused by discrete distance sampling: the correlation function for distances greater than or equal to 1km and less than 3km is averaged as the correlation function for a distance of 2km; the correlation function for distances greater than or equal to 2km and less than 4km is averaged as the correlation function for a distance of 3km; and so on, until the average correlation function corresponding to the preset maximum distance is calculated;

[0058] Based on the data moving average processing, a polynomial regression model is fitted between the wind speed observation correlation function r′(d) and the distance d. By comparing the squared errors between the correlation function values ​​calculated by the polynomial fitting models of different orders and the actual correlation function values, the optimal order and optimal coefficients are determined to minimize the sum of squared errors, thus obtaining the optimal polynomial fitting regression model r′(d) = a0 + a1×d + a2×d. 2 +...+a m ×d m (hereinafter referred to as Formula 1.2), where m is the order of the polynomial, and a0, a1, a2, ... a m These are the polynomial coefficients.

[0059] As a preferred option, the accuracy of the established optimal polynomial regression model is also verified, including calculating the coefficient of determination R. 2 Statistical indicators such as root mean square error (RMSE) are used to test the model's fitting accuracy and predictive ability. If necessary, the model can be optimized by adjusting the polynomial order or introducing regularization methods to ensure that the regression model can accurately describe the quantitative relationship between distance and correlation function within the evaluation area, providing a reliable mathematical basis for subsequent standardized observation random mean square error estimation and representativeness assessment.

[0060] S400. Estimation of the standardized random mean square error of observations:

[0061] Under the condition that the target assessment area meets the homogeneity and isotropy of the same climate zone, an analytical relationship is established between the wind speed observation correlation function r′(d) and the theoretical true value correlation function r(d). Based on this, the standardized observation random mean square error is calculated at d=0 using the intercept a0 of the regression model. or v It is used to measure the relative level of random error in regional wind speed observations.

[0062] Specifically, in calculating the standardized observation random mean square error or v At the same time, it should include at least the following sub-steps:

[0063] Assume the wind speed at point A is F(A), and its deviation and variance are respectively expressed as... f (A) and D v (A) indicates:

[0064] (1.3)

[0065] (1.4)

[0066] For meteorological elements within the same climate zone, the average random error of their observations is almost zero, therefore the average of their observed values ​​is... and the average of the true values They are equal. Assume the observed values ​​at point A are equal. Equals the true value of point A, F(A), plus observational random error. d v (A), then the observed value of point A The deviation contains two components, one being the true value of the feature. deviation f (A) The other is observational random error. d v (A), that is:

[0067] (1.5)

[0068] (1.6)

[0069] here d v (A) Dimensions and f (A) Consistency, also known as random observation error. According to d vThe randomness of (A) is such that, since the observational random errors at different points (A≠B) are uncorrelated, the observational random error at a certain point is also uncorrelated with the deviations at that point or other points. That is:

[0070] (1.7)

[0071] (1.8)

[0072] (1.9)

[0073] When two points coincide (A=B): (1.10)

[0074] in Let $A$ be the mean square value of the random error observed at point A, also known as the random mean square error of observation; then the standardized random mean square error of observation... or v (A), also known as relative observation random mean square error, where:

[0075] (1.11)

[0076] The correlation function calculated based on the observation data is as follows:

[0077] (1.12)

[0078] Expanding equation (1.12), and substituting the results from equations (1.6)-(1.10), equation (1.12) transforms into:

[0079] (1.13)

[0080] Then, equation (1.13) is transformed into:

[0081] (1.14)

[0082] Substituting equations (1.1) and (1.11) into equation (1.14), we get:

[0083] (1.15)

[0084] Assuming the correlation function of wind speed parameters satisfies homogeneity and isotropy within the same climate zone of the assessment area, then the correlation function is solely a function of distance. Furthermore, the standardized observational random mean square error for each point within the region is also considered. or v If they are all equal, then assuming the distance between A and B is d, equation (1.12) can be simplified to:

[0085] (1.16)

[0086] because or v Always positive, the above formula shows the correlation function of observations at the same distance. The correlation function must be less than the theoretical truth value. r v (d) means that the correlation between theoretical true values ​​and observed values ​​is better at the same distance. When d=0, stations A and B coincide, and the theoretical true values... r v Substituting (d)=0 into equation (1.16) and making a transformation, we can obtain... (hereinafter referred to as Equation 1.17), where To calculate the correlation function for the observed data, substitute d=0 into the fitted wind speed correlation function and distance regression model, i.e., equation (1.2), to obtain... (Hereinafter referred to as Equation 1.18). Substituting Equation (1.18) into Equation (1.17) yields... or v =1 / a0-1 (hereinafter referred to as Equation 1.19). Therefore, it can be seen that the relationship between the correlation function and distance, derived through a polynomial fitting regression model, can be used to calculate... Or a0, thus obtaining the standardized observation random mean square error. or v . or v This constitutes the normalized ratio of the random mean square error of observation to the variance of the elements, which is used to measure the uniform weakening effect of observation noise on correlation and its benchmark contribution to the representativeness assessment threshold system.

[0087] S500. Calculation of substitution error for reference stations:

[0088] Based on the regression model and standardized observation random mean square error, the relative standard error generated by replacing the true wind speed value of the target point with the wind speed observation data of candidate reference stations in the evaluation area is calculated. E r .

[0089] Specifically, the relative standard error arising from substituting the wind speed at the reference station for the wind speed at the target point. E r Its implementation includes at least the following sub-steps:

[0090] S501. Based on the theoretical truth correlation function method, the relative standard error generated by replacing the data of target point B, which is a distance d away from a certain reference station (A), with the data of a certain reference station (A) is calculated. E r .

[0091] Because the observation data from reference station A includes both theoretical true values ​​and observational random errors, let's assume that using the observation data from station A to replace the theoretical true values ​​from station B will result in a standardized bias. E r for:

[0092] (1.20)

[0093] In the formula: S(A) and S(B) are the standardized deviations of the theoretical true values ​​of sites A and B, respectively; m (A) is the standardized random error of the observation at point A. The expressions for the three are as follows:

[0094] (1.21)

[0095] (1.22)

[0096] (1.23)

[0097] in: d (A) represents observational random error. or v To standardize the random mean square error of observations, f (A) and D v (A) represents the wind speed deviation and variance at point A, respectively. f (B) and D v (B) represents the wind speed deviation and variance at point B, respectively.

[0098] Expanding equation (1.20) yields:

[0099] (1.24)

[0100] From the results of equations (1.7)-(1.10), we can obtain:

[0101] (1.25)

[0102] (1.26)

[0103] (1.27)

[0104] (1.28)

[0105] Substituting the results of equations (1.25)-(1.28) into equation (1.24), we get:

[0106] (1.29)

[0107] in: (1.30), (1.31)

[0108] Assuming the correlation function between stations A and B takes into account the effect of distance, then:

[0109] (1.32)

[0110] Substituting equations (1.30)-(1.32) into (1.29), we get:

[0111] (1.33)

[0112] Where: r(d) is the theoretical truth correlation function for any two stations A and B with a distance of d. or v This is to standardize the random mean square error of observations.

[0113] S502. Calculate the relative standard error of wind speed data from the reference station (A) instead of the target point (B) based on the correlation function method of actual observations. E r .

[0114] As can be seen from equation (1.33) in step S501, the error caused by the substitution of the observation data of the reference station A for the target point B comes from two aspects: one is the random error brought about by the observation of the station data itself. or v Another error term is due to data changes caused by the distance d relationship, which is 2-2. r (d) The magnitude of this error is related to the representativeness of the reference stations. If it does not exceed the random error... or v That is, (2-2) r (d))≤ or v A positive result indicates good representativeness; otherwise, the representativeness is average or poor. This formula also shows that as long as the correlation function of wind speed within the assessment area varies with distance, the relative standard error of wind speed data representing any target point at a distance *d* from the reference station can be determined. E r However, the relevant functions here are only theoretical values.

[0115] The wind speed correlation function of the theoretical truth can be obtained using equation (1.16). r v (d) The relationship between the observed wind speed at the same distance d and the function of the correlation function (1.16) can be obtained by transforming this equation:

[0116] (1.34)

[0117] Substituting (1.17) and (1.34) into (1.33), we finally obtain the relative standard error. E r as follows:

[0118] (1.35)

[0119] In the above formula , The correlation functions for zero and d distances are respectively, and can be calculated by fitting a multinomial regression model of the correlation function with respect to distance.

[0120] S600. Representativeness Level Classification:

[0121] Based on standardized observation random mean square error or v and the relative standard error of the reference station E r The representativeness of the reference stations is divided into several levels according to the threshold range, and each level corresponds to different engineering application suggestions.

[0122] Specifically, based on standardized observation random mean square error or v Data bias caused by distance between stations 2-2 r (d) Not greater than the standardized observation random mean square error or v That is, (2-2) r (d))≤ or v Based on the principle that the representativeness is considered good or above, the relative standard deviation of wind speed represented by the reference station due to spatial distance can be divided into four error levels. The threshold for the relative standard deviation and the application suggestions for the reference station are shown in Table 1 below, which are as follows: Level I Excellent. E r ≤ 1.5 or v It can be directly used for engineering parameter calculation and climate analysis; Level II (Good), 1.5 or v < E r ≤ 2 or v Used for routine climate analysis and general engineering parameter estimation; caution should be exercised when using it for extreme event analysis; Level III qualified, 2 or v < E r ≤ 2.5 or vThis is only a rough estimate and needs to be corrected with other data; Level IV, E r >2.5 or v This cannot be used as a reference point. or v It can be obtained from equation (1.17) or equation (1.19).

[0123] Table 1. Classification Criteria and Application Recommendations for Representative Meteorological Reference Stations

[0124]

[0125] S700. Spatial Distance Inversion and Representativeness Assessment:

[0126] The spatial distance range corresponding to each level threshold is calculated by inversion, thereby enabling the representativeness and quality of the spatial range of the reference station to be judged and evaluated, and providing a basis for the selection of reference stations for engineering applications with different accuracy requirements.

[0127] Specifically, the relative standard deviation of any point at a distance d within the calculated evaluation area will be used as the basis for the evaluation. E r The calculation formula (1.35) is transformed into:

[0128] (1.36)

[0129] The relative standard deviation of the representativeness level classification of meteorological reference stations E r Substituting the threshold into equation (1.36), and combining it with the regression model of the correlation function and distance obtained from the fitting (Equation 1.2), the spatial distance range corresponding to the representative level of the meteorological reference station can be calculated. This enables the excellent discrimination and evaluation of the representativeness of the spatial range of the wind speed meteorological reference station, and provides a basis for the selection of reference stations for engineering applications with different accuracy requirements.

[0130] Example 2: Case Verification

[0131] Based on the above embodiment 1, in order to better understand the present invention, the following embodiment 2 is used to conduct a representative assessment of wind speeds at reference stations in four cities in southern Hebei Province (Shijiazhuang, Hengshui, Xingtai and Handan) for each season to further clarify the content of the present invention.

[0132] In its specific implementation, this invention first collects station information and daily average wind speed data from the preprocessing assessment area (southern Hebei), selects independent samples to calculate the correlation function and distance between each pair of stations, and uses the wind speed correlation function to calculate the standardized observation random mean square error. Then, it calculates the relative standard error caused by the wind speed of the reference station replacing the target point's wind speed using the relationship between the correlation function and distance. Based on the principle that the error caused by the reference station replacing the target point due to distance factors does not exceed a certain standardized observation random mean square error threshold, the representativeness level of the reference stations is classified, and finally, the representativeness of the reference stations is assessed. The method of this invention includes:

[0133] S100. Collection and preprocessing of site topography and daily average wind speed data.

[0134] Environmental information and average wind speed data were collected from 331 meteorological stations in four cities in southern Hebei Province (Shijiazhuang, Hengshui, Xingtai, and Handan) within the assessment area. This included the longitude, latitude, altitude of each station, and the required daily average wind speed data for the six years from 2018 to 2023. The raw wind speed observation data were then preprocessed. A climate threshold check was used to remove spurious values ​​exceeding the climate threshold. A spatiotemporal consistency check was performed to identify and remove dead or erroneous values. Correlation analysis was used to remove 39 stations whose correlation coefficients with more than half of the stations within a 25km radius failed the 0.1 confidence level test. A reliable wind speed database of 292 stations (61 national stations and 231 provincial stations) was established.

[0135] S200. Select 50 or more independent samples from each station in the region during the same time period of the four seasons.

[0136] Based on the wind speed meteorological observation data preprocessed in step S100, to ensure the independence of the selected meteorological element samples, avoid the influence of weather system cycles, and guarantee that the number of samples for each evaluation season is more than 50, the data sample selection method is as follows: For winter (January, February, and December), spring (March, April, and May), summer (June, July, and August), and autumn (September, October, and November), the first day of each ten-day period of each month is selected. If there is missing data, the next day is selected. Nine samples can be selected for each season of the year, obtained from 2018 to 2023, for a total of 54 samples per quarter.

[0137] S300~S400. The standardized observation random mean square error is calculated using the wind speed correlation function.

[0138] Based on step S200, 54 independent samples were selected from each station in the evaluation area for each season, and the correlation function between each pair of stations was calculated. The calculation formula is as follows: ,inf (A) and f (B) represents the deviation of wind speed values ​​at stations A and B, respectively. D v (A) and D v (B) represents the variance of wind speed at stations A and B, respectively, and “—” indicates the average of the time series.

[0139] Based on N=42486 sets of data obtained from the station distances and corresponding correlation function fields, scatter plots of correlation functions corresponding to different distances in each season were drawn (see...). Figure 2 to Figure 5 (blue dot); then perform a 3km moving average: average the correlation functions for distances greater than or equal to 1km and less than 3km, and use this as the correlation function for a distance of 2km; average the correlation functions for distances greater than or equal to 2km and less than 4km, and use this as the correlation function for a distance of 3km; and so on, to calculate the average correlation functions for different distances in each season for distances of 2, 3, 4...200km (see...). Figure 2~Figure 5 (The red moving average curve).

[0140] A multinomial regression model is fitted to the correlation function and distance based on the moving average of each distance correlation function. By comparing the squared errors between the correlation function values ​​calculated from polynomial fitting models of different orders and the actual correlation function values, the optimal order and best coefficients are determined to minimize the sum of squared errors. This allows for the fitting of binomial regression models for each season, i.e.:

[0141] winter:

[0142] spring:

[0143] summer:

[0144] autumn:

[0145] in The correlation function is the average wind speed observation value, where d is the distance between stations, and the goodness of fit of the binomial regression model for each season is the coefficient of determination R. 2 All values ​​are greater than 0.99 and close to 1.0, indicating a very good fit.

[0146] The intercept of the binomial regression model based on the fitted correlation function and distance is its constant term a0, from which the standardized observation random mean square error can be calculated. or v ,Right now or v=1 / a0-1. The intercept a0 and standardized observation random mean square error of the polynomial regression model for each season within the evaluation region. or v See the table below:

[0147] Table 2. Intercept a0 and standardized observation random mean square error of the binomial regression model for each season. or v

[0148]

[0149] S500. The relative standard error arising from substituting the wind speed at the reference station for the wind speed at the target point. E r .

[0150] The relative standard error arising from the substitution of wind speed observation data from reference station A in the evaluation area for the true wind speed data of target point B, which is located at a distance d from A, is as follows. E r Theoretical derivation yields Where r(d) is the theoretical truth correlation function with distance d. or v This is to standardize the random mean square error of observations.

[0151] The relative standard error generated by substituting wind speed data from target point B with wind speed observation data from reference station A based on the correlation function method of actual observation values. E r The wind speed correlation function can be obtained using the theoretical true value. r v (d) Correlation function with observed wind speed at the same distance d From the relationship, we can deduce that:

[0152]

[0153] In the above formula , The correlation functions for zero and d distances are respectively obtained by fitting a multinomial regression model of the correlation function and the distance. Figure 6 It assesses the relative standard error of wind speed in southern Hebei during different seasons. E r A graph showing how the distance d changes. Figure 6 It can be seen that the relative standard error of the winter and spring seasons is smaller than that of the summer and autumn seasons.

[0154] S600. Determine the criteria for classifying the representativeness of wind speed at the reference stations as excellent or poor.

[0155] Based on the calculated standardized observation random mean square error or vThe wind speed at the reference stations can represent the relative standard deviation caused by spatial distance. E r It is divided into four levels: Representative Level I (Excellent), E r ≤ 1.5 or v It can be directly used for engineering parameter calculation and climate analysis; its representativeness level is good (Level II), 1.5. or v < E r ≤ 2 or v Used for routine climate analysis and general engineering parameter estimation; caution should be exercised when using it for extreme event analysis; representativeness level III qualified, 2 or v < E r ≤ 2.5 or v This is only a rough estimate and needs to be corrected with other data; representativeness level IV. E r >2.5 or v This cannot be used as a reference site. Table 3 below shows the relative standard deviation of the four representative levels for each season in the southern Hebei region of the assessment area. E r Threshold.

[0156] Table 3. Representativeness Level and Distance Assessment of Wind Speed ​​at Reference Stations

[0157]

[0158] S700. Enables representative assessment of wind speed meteorological reference stations.

[0159] The relative standard deviation of any point representing a distance d from the meteorological reference station is calculated. E r (See Figure 6 Then, based on the relative standard deviation of the representative level classification standard for wind speed spatial distance of the reference stations, E rBy setting thresholds, the representative spatial distance ranges for each level can be obtained (see Table 3). Based on the distance ranges given in Table 3, the representativeness level assessment of wind speed at meteorological reference stations in southern Hebei can be achieved. For reference stations analyzing annual wind speed data, the principle of small errors in each season must be met. Therefore, meteorological reference stations with a distance of less than or equal to 13 km from the target point are optimal in southern Hebei and can be used directly. Meteorological reference stations with a distance greater than 46 km from the target point cannot be used as reference stations for analyzing annual wind speed data because they do not meet the winter level requirements; however, they can be used for analyzing summer and autumn wind speed data.

[0160] The objectives of this invention have been fully and effectively achieved through the above embodiments. Those skilled in the art will understand that this invention includes, but is not limited to, the contents described in the accompanying drawings and the specific embodiments described above. Although the invention has been described with reference to what is currently considered the most practical and preferred embodiments, it should be understood that the invention is not limited to the disclosed embodiments, and any modifications that do not depart from the functional and structural principles of the invention will be included within the scope of the claims.

Claims

1. A method for representative evaluation of a regional wind speed meteorological reference station based on a correlation function field, characterized in that, At least comprising the following steps: S100. Collecting geographical environmental information and multi-year daily wind speed observation data of high-density meteorological observation sites belonging to the same climate zone and having wind speed observation in the target evaluation area, and performing quality control and preprocessing on the collected wind speed data to form a wind speed element sample database; S200. Based on the preprocessed wind speed data, for the daily wind speed observation data of each meteorological observation site in the target evaluation area, selecting independent samples in each preset evaluation period with the same time sequence and at a set time interval to form a wind speed element sample set with statistical independence; S300. Based on the constructed wind speed element sample set, statistical distance and wind speed observation correlation function between all sites are calculated to obtain the relationship field of site distance and corresponding correlation function; The relationship field is processed by moving average according to distance to obtain the average correlation function value corresponding to each distance segment, and a regression model of wind speed observation correlation function r'(d) with respect to site distance d is established; S400. Establish the analytical relationship between the wind speed observation correlation function r'(d) and the theoretical true value correlation function r(d), and accordingly calculate the standardized observation random mean square error at d = 0 from the intercept a0 of the regression model η v ; S500. Based on the regression model and the standardized observation random mean square error η v , the relative standard error of the estimated wind speed observation data of the candidate reference station in the evaluation area instead of the true value of the target point wind speed is calculated E r ; S600. Random mean square error based on standardized observations η v and reference station substitution relative standard error E r , the reference station representative is divided into several grades from good to bad according to the threshold interval, wherein when the regression model of the wind speed observation correlation function r'(d) about the distance d between the stations is constructed, at least includes: Based on the N groups of data in the relationship field composed of the spatial distance between two sites and the corresponding wind speed observation correlation function, 3km interval moving average processing is performed: the correlation functions with distance greater than or equal to 1km and less than 3km are averaged as the correlation function corresponding to a distance of 2km; the correlation functions with distance greater than or equal to 2km and less than 4km are averaged as the correlation function corresponding to a distance of 3km; and so on until the average correlation function corresponding to the preset maximum distance is calculated; Based on the data moving average processing, a polynomial regression model is fitted between the wind speed observation correlation function r′(d) and the distance d. By comparing the squared errors between the correlation function values ​​calculated by the polynomial fitting models of different orders and the actual correlation function values, the optimal order and optimal coefficients are determined to minimize the sum of squared errors, thus obtaining the optimal polynomial fitting regression model r′(d) = a0 + a1×d + a2×d. 2 +…+a m ×d m Where m is the order of the polynomial, a0, a1, a2, ... a m These are the polynomial coefficients; S700. Inverse calculation of the spatial distance range corresponding to each representative grade threshold value, thereby realizing the representative and excellent discrimination and evaluation of the spatial range of the reference station in the evaluation area.

2. The method of claim 1, wherein the method further comprises: In the above step S100, the collection and preprocessing of the site geographical environmental information and wind speed observation data in the target evaluation area at least comprises: S101. Determining all meteorological sites in the target evaluation area belonging to the same climate zone and having wind speed observation, and the spatial distribution density of which meets the requirement that the distance between adjacent sites is not greater than 25km; S102. Collecting geographical environmental information and wind speed daily observation data of each meteorological site, at least including the longitude, latitude, altitude of the site, and wind speed daily observation data for more than 6 years; S103. Preprocessing the collected wind speed observation data of each site, using a climate threshold checking method to eliminate false values exceeding the climate threshold in the original data, and identifying and eliminating the dead values or wrong values through spatial and temporal consistency checking; S104. Performing spatial correlation statistical test on the preprocessed wind speed data of each site, calculating the correlation coefficient of the wind speed observation data of each site with all sites within a 25km range, and eliminating the sites whose correlation coefficients with half of the sites within a 25km range cannot pass the 0.1 significance test; S105. Establishing a wind speed element sample database based on the sites and their wind speed observation data that have completed quality control and correlation test, at least including site geographical environmental information and daily wind speed observation data.

3. The method of claim 1, wherein the method further comprises: In the step S200, when selecting the independent samples of the preset evaluation time period of each station in the target evaluation area, at least the following sub-steps are included: S201. According to the time variation characteristics of meteorological elements and engineering application requirements, the preset evaluation period is divided into three evaluation period types of month scale, season scale and year scale; S202. In order to ensure that the selected meteorological element samples have independence and avoid the influence of weather system cycle, an equal interval sampling strategy with fixed starting point and fixed step length is adopted, the minimum sampling interval is limited to weaken the influence of weather scale autocorrelation, and the same phase is repeatedly extracted on the multi-year sequence to ensure that the number of independent samples of each station in each evaluation period is not less than a preset threshold; S203. Based on the independent samples of each station in each evaluation period obtained by sampling, a wind speed element sample set with statistical independence is constructed, which at least includes station code, sample time identifier and wind speed observation value.

4. The method of claim 3, wherein the method further comprises: When sampling the three evaluation periods of month scale, season scale and year scale: For monthly period evaluation sampling, a strategy of selecting a group of samples every 3 days is adopted, if missing data is encountered, the next day is taken in sequence, ensuring that at least 9 groups of independent samples are obtained each month, and the same phase is repeatedly extracted on the multi-year sequence until at least the minimum sample number required for statistical analysis is obtained; For seasonal period evaluation sampling, the year is divided into four seasonal periods of winter, spring, summer and autumn, a strategy of selecting a group of samples every 10 days is adopted, if missing data is encountered, the next day is taken in sequence, ensuring that at least 9 groups of independent samples are obtained each season, and the same phase is repeatedly extracted on the multi-year sequence until at least the minimum sample number required for statistical analysis is obtained; For annual period evaluation sampling, a strategy of selecting a group of samples on the same date each month is adopted, ensuring that at least 12 groups of independent samples are obtained each year, and the same phase is repeatedly extracted on the multi-year sequence until at least the minimum sample number required for statistical analysis is obtained.

5. The method of claim 1, wherein the method further comprises: In the step S300, when constructing the distance relationship field between stations in the target evaluation area and the corresponding correlation function, at least the following steps are included: Based on the collected geographical environment information of each station, the spatial distance between each two stations is calculated to obtain distance data of N=k×(k-1) / 2 station pairs, where k is the number of stations; Based on the selected evaluation period data sample, based on The correlation function of wind speed observation between any two stations in the region is counted, wherein A and B represent any two stations in the region, r v (A, B) is the correlation function of wind speed observation between stations A and B, which represents the correlation degree of wind speed observation data between the two stations, and d is the distance between the stations, and D v (A) and D v (B) are the wind speed variances of stations A and B, respectively, and the average of the time series is represented by "--". The calculated spatial distance between each two stations and the corresponding wind speed observation correlation function are paired and combined to form the distance relationship field between stations and the corresponding correlation function.

6. The method of claim 1, wherein the method further comprises: The step S400 includes: Based on the assumption that the wind speed factor in the same climate zone approximately satisfies the uniformity and isotropy, the analytical relationship between the observed correlation function r'(d) and the theoretical true correlation function r(d) is established as follows: r'(d) = r(d) / (1 + 2 η v ) At spatial distance d = 0, r'(0) is estimated from the intercept a0of the regression model of the wind speed observation correlation function r'(d), r'(0) being the value of the regression model at zero distance, from which the normalized observation random mean square error η v = 1 / a0- 1.

7. The method of claim 1, wherein the method further comprises: In the step S500, the relative standard error of the wind speed at the reference station instead of the wind speed at the target point is calculated E r The regression fitting result of the wind speed observation correlation function is taken as input and the calculation is performed according to E r =1+[1-2·r′(d)] / r′(0) is calculated, where r′(d) is the value of the regression model at a distance d, and r′(0) is the value of the regression model at zero distance.

8. The method of claim 1, wherein the method further comprises: In step S600, the principle that the data difference caused by distance does not exceed the regional observation random error is used to standardize the observation random mean square error η v The reference threshold is set as E r The threshold interval is divided into multiple levels, and the acceptability of the error caused by the reference station replacing the target point at different distances is determined by grading, and each level corresponds to different engineering application suggestions, and finally a reference station wind speed representative excellent grading standard is formed, wherein: the first-class excellent, E r ≤ 1.5 η v , which can be directly used for engineering parameter calculation and climate analysis; the second-class good, 1.5 η v < E r ≤ 2 η v , which is used for general climate analysis and engineering parameter estimation, and extreme event analysis should be used with caution; the third-class qualified, 2 η v < E r ≤ 2.5 η v , only for rough estimation and needs to be corrected with other data; the fourth-class poor, E r > 2.5 η v , which cannot be used as a reference station.

9. The method of claim 1, wherein the method further comprises: In the step S700, the spatial distance inversion and the representative evaluation of the wind speed meteorological reference station are realized according to the above, at least including: substituting the threshold value of the representative grade of each meteorological reference station into the wind speed observation correlation function r'(d) and the regression model between the spatial distance d E r The transformation relationship r'(d)=0.5·[1+r'(0)-r'(0)· E r ], wherein r'(0) is the value of the regression model at zero distance, combining the regression model between the fitted wind speed observation correlation function r'(d) and the spatial distance d, the spatial distance range corresponding to the representative grade of each meteorological reference station is inversely calculated, and the representative good discrimination and evaluation of the spatial range of each wind speed meteorological reference station in the target evaluation area are realized.

10. A computer program product comprising computer instructions, characterized in that, The computer instructions are used to execute the representative evaluation method of regional wind speed meteorological reference station based on correlation function field according to any one of claims 1-9.

11. A computer-readable storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to implement the representative evaluation method of regional wind speed meteorological reference station based on correlation function field according to any one of claims 1-9.

Images