A method for processing near-surface annual maximum wind speed data into a time scale by considering space-time correction

Abstract

The application discloses a kind of near-surface annual maximum wind speed data reduction time scale processing methods considering space-time correction, it is related to extreme wind speed data processing field.Solve the urgent need for a kind of wind speed spectrum correction method in prior art, which can process ERA5-Land reanalysis wind speed data high-frequency information loss, realize the accurate conversion of hour-level data to 10min scale extreme wind speed, and give consideration to cost and efficiency.The method comprises the following steps: for any grid point, calculate the average wind speed of different seasons in the region, and obtain the fluctuating wind speed sequence by de-meaning processing; estimate the power spectral density of fluctuating wind speed using Welch average periodogram method; based on the power spectral density, calculate the zeroth moment and second moment; reconstruct the modified mixed wind speed spectrum; based on the mixed wind speed spectrum, correct the spectral moment and the correction peak factor; and calculate the corrected annual maximum expected wind speed; compare the corrected annual maximum expected wind speed with the observed annual maximum wind speed of the original sample sequence, and output the modified 10min annual maximum wind speed.

Description

Technical Field

[0001] This invention relates to the field of extreme wind speed data processing technology, specifically to a method for reducing the time scale of near-surface annual maximum wind speed data considering spatiotemporal correction. Background Technology

[0002] Against the backdrop of global climate change and the accelerated transition to renewable energy, the accurate characterization and quantification of extreme wind speeds are crucial for fundamental meteorological research, structural engineering design, and wind energy resource assessment. In the engineering design of critical infrastructure such as wind power generation, high-rise buildings, roads and bridges, and ultra-high-voltage transmission lines, extreme wind speeds directly determine the wind load, fatigue life, and service life of the structure. Basic wind speed is a core parameter for wind energy utilization and wind-resistant design of buildings and infrastructure. From a research perspective, the study of basic wind speed can be based on observational data and ERA5-Land reanalysis data. Due to the uneven spatial distribution of ground meteorological stations and issues such as station relocation, analysis using high-resolution ERA5-Land reanalysis data is of significant value. The spatial resolution of ERA5-Land reanalysis data is 0.1° × 0.1°, and the temporal resolution is hours. Basic wind speed is defined as: "the wind speed determined by statistically analyzing the 10-minute average wind speed observation data at a height of 10m on a local open, flat surface, based on a 50-year return period maximum value." Clearly, using ERA5-Land reanalysis data requires reducing the time scale of the maximum wind speed in the analysis data.

[0003] In wind-resistant engineering design, the 10-minute annual maximum wind speed is a core design parameter, and its accuracy directly affects the wind resistance safety and economy of the engineering structure. Currently, one of the main data sources for obtaining large-scale, long-term wind speed data is ERA5-Land reanalysis data. Due to its wide coverage and complete time series, ERA5-Land reanalysis data is widely used in related research and practice.

[0004] However, existing technologies for estimating extreme wind speeds using ERA5-Land reanalysis data face numerous technical challenges that urgently need to be addressed, as follows: First, the ERA5-Land reanalysis wind speed data has an hourly time resolution. This resolution limitation results in the absence of high-frequency information of small-scale turbulence in the data. Its original power spectral density cannot truly reflect the turbulent characteristics of the wind field and has obvious distortion, which in turn affects the accuracy of subsequent calculations of related parameters based on the power spectrum.

[0005] Secondly, the original spectral moments and peak factor estimations of ERA5-Land reanalysis wind speed data have biases. Existing technologies cannot accurately convert hourly ERA5-Land reanalysis data into the 10-minute scale annual maximum wind speed required for engineering wind resistance design. As a result, ERA5-Land reanalysis data cannot be directly applied to engineering wind resistance extreme value design, making it difficult to meet actual engineering needs.

[0006] Finally, when using the original ERA5-Land reanalysis data directly to estimate extreme wind speeds, the aforementioned defects lead to large deviations in the estimation results, which cannot provide reliable data support for the wind-resistant design of engineering structures. This may affect the wind-resistant safety of engineering structures and pose potential engineering risks.

[0007] Therefore, acquiring large-scale, long-term extreme wind speed data on a 10-minute timescale is fundamental to supporting disaster prevention and mitigation in engineering projects and the development of the wind power industry. Addressing the shortcomings of existing technologies, there is an urgent need for a wind speed spectrum correction method that can resolve the lack of high-frequency information in ERA5-Land reanalysis data, achieve accurate conversion of hourly data to extreme wind speeds on a 10-minute timescale, and balance cost and efficiency, in order to meet the practical needs of wind engineering design. Summary of the Invention

[0008] This invention addresses the urgent need in existing technologies for a wind speed spectrum correction method that can resolve the lack of high-frequency information in ERA5-Land reanalysis wind speed data, achieve accurate conversion of hourly data to extreme wind speeds on a 10-minute scale, and balance cost and efficiency. To this end, this invention proposes a method for downscaling near-surface annual maximum wind speed data, considering spatiotemporal correction. The invention achieves this through the following technical solution: Option 1: This invention proposes a method for reducing the time scale of near-surface annual maximum wind speed data, considering spatiotemporal correction. The method includes the following steps: Step 1: Divide the study area into several wind speed zones based on the national basic wind speed distribution map or industry standard wind speed zoning map, and calculate the average wind speed in different seasons within each zone, taking into account the seasonal characteristics of wind speed in the zone. Step 2: For the hourly wind speed sequence U(t) within any grid point, use the average wind speed of different seasons within the region to which the grid point belongs. The mean-free processing was performed to obtain the fluctuating wind speed sequence. ; Step 3: Based on the fluctuating wind speed sequence obtained in Step 2 The Welch average periodogram method was used to estimate fluctuating wind speed. power spectral density ; Step 4: Based on power spectral density Calculate the zeroth and second moments; Step 5: Wind speed spectrum based on the original wind speed sequence Reconstructed and corrected mixed wind speed spectrum The mixed wind speed spectrum It includes a low-frequency band and a high-frequency band. The low-frequency band retains the original power spectral density pattern, while the high-frequency band adopts... Tail compensation is performed using the power law, where the starting frequency of the high-frequency band is a preset cutoff frequency and the ending frequency is the target frequency. Step 6: Based on the mixed wind speed spectrum Corrected spectral moments and correction peak factor; Step 7: Calculate the corrected annual maximum expected wind speed using the corrected spectral moments and correction peak factor; Step 8: Compare the corrected annual maximum expected wind speed with the observed annual maximum wind speed from the original sample sequence. Compare the results and output the corrected 10-minute annual maximum wind speed.

[0009] Furthermore, a preferred embodiment is provided in which step 1 further includes the step of taking into account the wind speed distribution pattern that takes into account spatial and seasonal influences, thereby achieving the step of de-averaging the accurate pulsating wind speed in step 2.

[0010] Furthermore, a preferred embodiment is provided in which the Welch average periodogram method is used to estimate the fluctuating wind speed in step 3. power spectral density The method is as follows:

[0011] Where T is the integral time scale, representing the average time of the turbulent vortex; The variance of the fluctuating wind speed. The cycle frequency.

[0012] Furthermore, a preferred embodiment is provided, wherein in step 4, the power spectral density is used as the basis for... The method for calculating the zeroth and second moments is as follows:

[0013]

[0014] In the formula, The power spectrum of the Gaussian process u(t) is given by angular frequency. , For filters, For averaging time, the original spectral moments are calculated from... =0 to The integration is capped at the Nyquist frequency and cutoff at the Nyquist frequency. An ideal low-pass filter.

[0015] Furthermore, a preferred embodiment is provided, wherein in step 5, the wind speed spectrum is based on the original wind speed sequence. Reconstructed and corrected mixed wind speed spectrum The method is as follows: Set cutoff frequency As the dividing point between the model spectrum and the physical spectrum, The original model spectrum is partially preserved; a target frequency is set. As the upper limit of integration, it corresponds to the Nyquist frequency on a 10-minute timescale.

[0016] The segment is the high-frequency tail portion, applied Power-law tail compensation is applied to shift the spectral lines from... Extending to the Nyquist frequency corresponding to 10 minutes .

[0017] Furthermore, a preferred embodiment is provided, wherein in step 6, the mixture of wind speed spectra is used as the basis for the process. The methods for correcting the spectral moments and the peak factor are as follows: The formula for the corrected spectral moment is:

[0018] in, Corrected wind speed variance , This is the corrected variance of the derivative; For the annual maximum wind speed event, an average number of exceedances is set. , The cycle is 1 year. For wind speed to exceed The occurrence rate of peak factors that have a significant impact on the annual maximum wind speed Umax. :

[0019] in, The 10-minute average annual maximum expected wind speed. This represents the average wind speed. The standard deviation of wind speed;

[0020]

[0021] Substituting the modified spectral moment formula into the peak factor formula yields the modified peak factor. :

[0022] Furthermore, a preferred embodiment is provided, wherein the method for calculating the corrected annual maximum expected wind speed in step 7 using the corrected spectral moments and the correction peak factor is as follows: .

[0023] Furthermore, a preferred embodiment is provided in which, in step 8, the corrected annual maximum expected wind speed is compared with the observed annual maximum wind speed of the original sample sequence. The method for comparing and outputting the corrected 10-minute annual maximum wind speed is as follows: .

[0024] Option 2: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described in Option 1.

[0025] Option 3: A computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the method described in Option 1.

[0026] The advantages of this invention are: This invention discloses a method and system for processing near-surface annual maximum wind speed drop timescale data considering spatiotemporal correction. This method effectively compensates for the lack of high-frequency information in ERA5-Land reanalysis data and improves the physical accuracy of the power spectrum. The method employs a hybrid wind speed spectrum reconstruction strategy of low-frequency preservation and high-frequency compensation: the low-frequency band retains the spectral type of the original ERA5-Land reanalysis data, fully utilizing its simulation accuracy for large-scale weather processes; the high-frequency band introduces the Kolmogorov turbulence power law for tail compensation, effectively restoring the small-scale turbulent energy distribution lost due to hourly time resolution limitations. This makes the corrected power spectral density closer to the turbulence characteristics of the measured wind field, solving the problem of high-frequency turbulence information distortion caused by directly using ERA5-Land reanalysis data in existing technologies.

[0027] The present invention describes a method for processing near-surface annual maximum wind speed timescale data that considers spatiotemporal correction. Based on the corrected mixed wind speed spectrum, the spectral moments are recalculated and the peak factor is corrected. An annual maximum expected wind speed estimation model adapted to the mixed spectrum is constructed, which corrects the extreme value estimation bias caused by the lack of high frequencies in the original data. The final output corrected 10-minute annual maximum wind speed conforms to the definition of 10-minute average wind speed in wind engineering design specifications, providing directly usable extreme wind speed parameters for wind engineering design and solving the pain point that ERA5-Land reanalysis data cannot be directly used for engineering design.

[0028] The present invention describes a method for processing near-surface annual maximum wind speed downscale data that considers spatiotemporal correction. This method does not rely on scarce long-term measured wind speed data and can complete the correction based solely on publicly available ERA5-Land reanalysis data. At the same time, it restores turbulence characteristics through spectral correction mathematical methods, avoiding the complex physical model calculations of traditional downscaling methods. While ensuring the correctness of the physical mechanism, it significantly reduces data acquisition costs and computational complexity, and improves the universality and efficiency of engineering applications.

[0029] The present invention describes a method for processing near-surface annual maximum wind speed timescale data that considers spatiotemporal correction. The corrected mixed wind speed spectrum takes into account both large-scale weather trends and small-scale turbulent fluctuations, making the results of subsequent spectral moment calculation, peak factor estimation, and annual maximum expected wind speed closer to the statistical characteristics of the real wind field. Compared with directly using the original ERA5-Land reanalysis data, the extreme wind speed estimation bias is significantly reduced, providing more reliable basic data support for the wind-resistant design of engineering structures and helping to improve the wind resistance safety of engineering structures.

[0030] This invention is also applicable to the engineering design of key infrastructure such as wind power generation, high-rise buildings, roads and bridges, and ultra-high voltage transmission lines. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating the time-scale reduction method for near-surface annual maximum wind speed data considering spatiotemporal correction as described in Implementation Method 1.

[0032] Figure 2 The near-surface 10m wind speed power spectral density of the ERA5-Land reanalysis and spectral correction method described in Implementation Method 1. With frequency A comparison diagram.

[0033] Figure 3 This is a schematic diagram of the probability distribution of the annual maximum wind speed before and after observation and spectral correction as described in Implementation Method 1.

[0034] Figure 4 This is a schematic diagram of the kernel density distribution of the annual maximum wind speed before and after observation and spectral correction as described in Implementation Method 1.

[0035] Figure 5 This is a schematic diagram of the time series of average annual maximum wind speed before and after the spectral correction method described in Implementation Method 1.

[0036] Figure 6 This is a schematic diagram of the time series of the annual maximum wind speed extreme value and mean value using the spectral correction method described in Implementation Method 1. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0040] Implementation Method 1, see Figures 1 to 6 This embodiment describes a method for downscaling near-surface annual maximum wind speed data by considering spatiotemporal correction. The method specifically includes the following steps: Step 1: Based on the national basic wind speed distribution map, divide the wind speed into zones. Different industries can use the corresponding basic wind speed distribution map or wind speed zoning standard as needed. For simplicity, this step uses the "Code for Wind Resistance Design of Highway Bridges" to divide the wind speed into five zones: Zone I (wind speed > 60 m / s); Zone II (50 m / s < 60 m / s); Zone III (40 m / s < 50 m / s); Zone IV (30 m / s < 40 m / s); and Zone V (wind speed ≥ 30 m / s). Combining the wind speed zoning with the seasonal characteristics of wind speed in each zone, calculate the average wind speed for each season within the zone. ; China is located in eastern Eurasia, on the western coast of the Pacific Ocean, and has a pronounced monsoon climate. In winter, prevailing winds are northerly, blowing from the continent towards the ocean, while in summer, prevailing winds are southerly, blowing from the ocean towards the land. The prevailing northerly winds and frequent cold waves in winter are closely related to the southward spread of cold air; the prevailing wind direction generally reflects the airflow direction under topographical conditions during cold air intrusion. The western and northeastern regions are located at different positions within the continental high-pressure zone and are influenced by topography, resulting in significant differences in prevailing wind directions. Northeasterly winds prevail in the northwest, while southwesterly winds prevail in central Northeast China. In summer, the prevailing wind direction near the surface is the opposite of that in winter, with southerly winds prevailing in most areas. The formation and changes of the southerly winds in summer coincide with the onset and northward movement of the East Asian summer monsoon.

[0041] In terms of average wind speed distribution, winds are strong in northern China and weak in the south; winds are strong along the coast and weak inland; winds are strong in plateaus and mountains and weak in basins and valleys. In addition, the maximum and average wind speeds in most parts of my country are sensitive to seasonal influences, with spring being the season with the highest wind speeds, followed by winter and summer, and autumn having the lowest wind speeds.

[0042] Because wind speeds vary significantly across different regions and seasons, the effects of different times and spaces should be considered during the preprocessing detrending process. The country should be divided into different locational regions based on the distribution of maximum wind speed, and further subdivided by time according to different seasons.

[0043] Step 2: For the hourly wind speed sequence within any grid point U(t) Calculate the average wind speed μ in this region during different seasons.i，j The mean-free processing was performed to obtain the fluctuating wind speed sequence. ; To quantify the energy distribution characteristics of wind speed fluctuations in the frequency domain and ensure that the spectral analysis focuses on the variance characteristics of turbulent fluctuations, the wind speed is demeaned and the hourly wind speed sequence is decomposed into mean flow and turbulent fluctuations.

[0044] For an hourly wind speed sequence U(t) within any grid point, first calculate the average wind speed μ of that region in different seasons. i，j The mean-free processing was performed to obtain the fluctuating wind speed sequence. .

[0045] (2-1) in, This is a zero-mean random process, and its variance represents the total energy of wind speed fluctuations. i represents different location zones, and j represents different seasons.

[0046] Step 3: Based on the fluctuating wind speed sequence obtained in Step 2 The Welch average periodogram method was used to estimate fluctuating wind speed. power spectral density ; Pulsating wind speed The energy characteristics are expressed through the power spectral density (PSD) function. Description. Power spectral density is the Fourier transform of the autocorrelation function of wind speed. For an ideal normal distribution, assuming the autocorrelation function of fluctuating wind speed decays exponentially with time, its one-sided power spectral density can be described by a theoretical formula: (2-2) Where T is the integral time scale, representing the average time of the turbulent vortex; The variance of the fluctuating wind speed. The cycle frequency.

[0047] Estimating fluctuating wind speed using the Welch mean periodogram method power spectral density The study employs a default Hamming window and a 50% overlap rate to suppress spectral leakage and reduce variance. Compared to the Fast Fourier Transform (FFT), the Welch method significantly reduces the variance (noise) of spectral estimation by calculating the average periodogram through overlapping data segments, making it easier to accurately identify the spectral slope in the mid-to-high frequency bands.

[0048] raw power spectral density The calculation is as follows: (2-3) in, This represents the correction periodicity diagram for the k-th windowed data segment. The cycle frequency and sampling frequency are set to [value]. Based on the Nyquist sampling theorem, the original power spectral density is estimated using the Welch method. The cycle frequency range is 0~12. This corresponds to the hourly resolution of ERA5-Land data.

[0049] Step 4: Based on power spectral density Calculate the zeroth and second moments; In order to incorporate time-domain statistics ( , This relates to frequency domain characteristics, introducing the concept of spectral moments. The j-th order spectral moment... Defined as power spectral density The weighted integral.

[0050] (2-4) The zeroth moment m0 ​​and the second moment m2 are obtained by integration: (2-5) (2-6) It is the power spectrum of the Gaussian process u(t), with angular frequency Due to the time resolution given It is a filter. For averaging time, the original spectral moments are calculated from... =0 to The integral is given by an upper limit of the Nyquist frequency. To ensure the stability of the extreme value estimation, this study uses a cutoff frequency of . An ideal low-pass filter.

[0051] Based on Parseval's Theorem and the differential properties of stochastic processes, the integral of the power spectrum in the frequency domain equals the total energy (variance) of the signal in the time domain, and the derivative process... The power spectrum is Its variance is the second moment of the original spectrum, that is: , Introducing both into the incidence rate From the formula, we can obtain its expression based on spectral moments: (2-7) Based on the original power spectrum The original spectral moments are calculated and used as a benchmark to verify the effectiveness of the mean removal process. Based on formulas (2-5) and (2-6), the spectral moment calculation methods consider the one-sided power spectral density and the Nyquist frequency of hourly wind speed. The original zeroth moment and second moment for: (2-8) (2-9) Step 5: Reconstruct the corrected mixed wind speed spectrum This includes: retaining the original power spectral density in the low-frequency band; and adopting a spectral pattern in the high-frequency band. Tail compensation is performed using the power law. (2-10) Step 6: Based on the mixed wind speed spectrum Step 7: Calculate the corrected annual maximum expected wind speed, including the corrected spectral moment and correction peak factor. (1) Calculation of corrected spectral moments Based on the spectral moment calculation formulas (2-8) and (2-9), the upper limit of the integration interval is extended to the target frequency. : (2-11) Right now Corrected wind speed variance , This is the corrected variance of the derivative.

[0052] (2) Calculation of annual maximum wind speed Assuming the exceedance probability of the annual maximum wind speed follows a Poisson distribution, for a given wind speed threshold... Baseline period Within, the maximum annual wind speed does not exceed probability Represented as: (2-12) The cycle is 1 year. For wind speed to exceed The incidence rate. For a zero-mean Gaussian process, By wind speed and its time derivative The joint probability density function determines this. Assume... and Since they are independent and both follow a Gaussian distribution, their probability density functions can be expressed as: (2-13) Here Pulsating wind speed , Pulsating wind speed The standard deviation of represents the turbulence intensity. denoted as the standard deviation of the rate of change of wind speed.

[0053] The incidence rate can be derived from the Rice formula. : (2-14) in, yes The conditional probability.

[0054] For the annual maximum wind speed event, an average number of exceedances is set. For the annual maximum wind speed U max Peak factor with significant impact : (2-15) The 10-minute average annual maximum expected wind speed. This represents the average wind speed. Let be the standard deviation of wind speed. Substituting it into the equation, we get... (2-16) (2-17) Substituting the modified spectral moment formula (2-11) into the peak factor formula (2-15), we obtain the modified peak factor. : (2-18) According to formula (2-16), the spectral-corrected 10-minute annual maximum expected wind speed is... (2-19) Step 8: Compare the corrected annual maximum expected wind speed with the observed annual maximum wind speed from the original sample sequence. Compare the results and output the corrected 10-minute annual maximum wind speed.

[0055] Spectral-corrected expected annual maximum wind speed Compared with the observed annual maximum wind speed in the original sample sequence Compare the values ​​and retain the maximum wind speed. This ensures that while compensating for the high-frequency energy deficit in the spectral correction model, the actual extreme wind speeds already captured in the sample are not underestimated.

[0056] (2-20) Example 1: This example is used to further explain and illustrate Implementation 1. (1) Energy spectral tail correction By extracting wind speed spectrum diagrams from multiple cities in 1950 and 2024 before and after spectral correction, it can be clearly observed that the power spectrum of the original wind speed sequence (red dashed line) exceeds the model's cutoff frequency ( After that, it showed an overall linear decay trend, and The missing power spectrum indicates that the high-frequency extreme wind speeds in the original ERA5-Land reanalysis data only extend to the hourly scale. Spectral correction reconstructs the power spectrum shape of the high-frequency tail of wind speeds (blue solid line), following the -5 / 3 turbulence theory scaling law at the cutoff frequency, raising the spectral slope to less than -5 / 3, and effectively extending the high-frequency band of the spectrum to [the desired range]. This corresponds to a 10-minute timescale. The spectral correction method effectively supplements and improves the missing high-frequency spectral energy density in the frequency domain, compensating for the microscale turbulence variance dissipated by the numerical model due to spatiotemporal smoothing, thus truly reflecting the maximum wind speed extreme characteristics on a 10-minute timescale. (a)-(i) represent the wind speed spectra at multiple coordinates in 1950 and 2024, respectively, where the red dashed line and blue solid line represent the spectra before and after the spectral correction method, respectively, and f > 1d. -1 After spectral correction, the slope of the spectrum was effectively increased to -5 / 3 and extended to 72d. -1 .

[0057] (2) Fitting of wind speed after spectral correction Using the annual maximum wind speed observed at 163 meteorological stations across the country from 1995 to 2022 as the evaluation benchmark, an error analysis was conducted on the data accuracy before and after the spectral correction method. Figure 3 The probability density function (PDF) and cumulative distribution function (CDF) of the observed data and reanalysis data before and after correction. Figure 3 (a) Compared with the station observation data (black solid line), the original ERA5-Land reanalysis data shows a significant leftward shift in its probability density peak and a complete absence of the long-tail characteristics of the high-wind-speed range, indicating a systematic and severe underestimation of NSWS over the Chinese land area. In contrast, the NSWS density center after spectral correction shifts significantly to the right and highly coincides with the peak position of the observation data, but the peak value of the curve is lower than that of the observation data. Figure 3 (b) In the cumulative distribution curve of wind speed, the extension trend of the long tail of the corrected wind speed in the high wind speed range (>20 m·s⁻¹) is more consistent with the observed data. The CDF curve as a whole tends to be consistent with the station observations, while the CDF curve of the ERA5-Land reanalysis data is generally skewed to the left, with the high wind speed region <20 m·s⁻¹. This indicates that the spectral correction method overcomes the defect of underestimating extreme values ​​in the original ERA5-Land reanalysis data to a certain extent, and can effectively reconstruct the probability distribution shape of the strong wind tail and restore the accuracy of extreme wind speeds.

[0058] Figure 4The distribution of the scatter plots before and after spectral correction is compared with the annual maximum wind speed observed at the station. The shades of color indicate the cluster density of the data points. Figure 4 (a) The high-density regions of ERA5-Land are significantly clustered below the 1:1 diagonal, and the underestimation of wind speed is mainly concentrated in the range of 10~20 m·s⁻¹. When the observed wind speed exceeds 20 m·s⁻¹, the scatter points of ERA5-Land data are almost all below the diagonal and sparsely distributed, indicating that mesoscale simulated wind speed cannot effectively capture extreme wind speeds at small scales. Figure 4 (b) The high-density centers of the annual maximum wind speed after spectral correction are concentrated above the 1:1 diagonal of the kernel density center point, the overall center of gravity shifts upward, and the scattered points extend to the 20~40 m·s⁻¹ range, which objectively confirms the good reproducibility of the spectral correction method for extreme wind speeds.

[0059] (3) Extreme wind speed error analysis By comparing the data with the measured data from the stations, this study conducted an error analysis on the annual maximum wind speed before and after the spectral correction method. The results are shown in Table 1. The average deviation between the original ERA5-Land annual maximum wind speed and the station's annual maximum wind speed was -3.95 m / s, indicating a serious underestimation of extreme wind speeds. After the spectral correction method, the average deviation was 1.23 m / s, and the overall absolute error was reduced by 2.71 m / s, indicating that the spectral correction method effectively compensated for the underestimation of extreme high-frequency energy in the reanalysis data. The root mean square error after correction decreased from 5.44 m / s to 5.36 m / s, effectively ensuring the robustness of wind speed quality while improving the extreme wind speed. The correlation coefficients between the data before and after correction and the measured data were 0.38 (p < 0.01) and 0.39 (p < 0.01), respectively, effectively reproducing the fluctuation trend of the maximum wind speed.

[0060] Table 1. Analysis of error parameters between the corrected spectral data and the annual maximum wind speed at the stations.

[0061] (4) Trend of annual maximum wind speed after spectral correction Figure 5 This study presents the long-term evolution of the national average annual maximum wind speed before and after spectral correction in the ERA5-Land reanalysis data from 1950 to 2024. Overall, the corrected 10-minute average annual maximum wind speed shows a significant leap compared to the original hourly average, with a regional average increase of approximately 6.8 m / s, effectively and directly compensating for the systematic underestimation of extreme wind speeds in the reanalysis data. Regarding long-term trends, both the original and corrected data exhibit an overall fluctuating upward trend, with the national average annual maximum wind speed after spectral correction showing an interannual increase rate of 0.005 m / s. -1 ) / a (p=0.005), with an interdecadal variation trend of 0.005 (ms). -1) / a (p=0.001). In addition, the 10-year moving average curve also reflects that, under the overall upward trend of wind speed, extreme wind speeds also have significant multi-decadal fluctuation characteristics. The extreme wind speeds in the late 1960s and 1990s were at a relatively low point, while the wind speeds in the 1980s and around 2010 entered a relatively high range. This further highlights the non-stationarity of the long-term evolution of extreme wind speeds.

[0062] Figure 6 This study presents the temporal evolution of the characteristic values ​​(national average, maximum, 99th percentile, and 95th percentile) of the national annual maximum wind speed after spectral correction from 1950 to 2024. Overall, different indicators of the long-term linear trend of wind speed have shown a consistent upward trend over the past 75 years, indicating an overall increase in near-surface extreme wind speeds in my country. Specifically, the national average series exhibits a significant long-term linear upward trend. Its 10-year moving average curve clearly reveals the multi-decadal non-stationary fluctuation process, namely, the average wind speed reached a decadal peak in the mid-to-late 1980s, then experienced a significant decline in the late 1990s, and resumed a steady upward trend around 2010. The long-term linear upward slope of the maximum value series is relatively gentle. This may be due to the sensitivity of the single-point absolute maximum value to occasional extreme weather events, resulting in the most dramatic interannual oscillations in this series. For example, around 1984, the series exhibited an abnormal peak of nearly 60 m / s, and the large interannual variance to some extent smoothed out the long-term linear fitting trend. The 99th quantile sequence, a key indicator characterizing the tail characteristics of extreme wind events, exhibits a significant long-term linear upward trend. This sequence accurately captures extreme wind speeds in specific years such as 1984, and its 10-year moving average trajectory fully reproduces the multi-decadal fluctuation pattern of "first rising, then falling, then rising again," consistent with the national average. The 95th quantile sequence is less affected by single extreme anomalies, and its long-term upward trend is the most robust and smooth. The continuous upward trend of this sequence further indicates that the baseline level of extreme wind speeds in most parts of the country is undergoing an overall increase, maintaining a significant strengthening trend over the past decade. Combining the changes in these four statistical indicators, it can be concluded that the long-term strengthening of extreme wind speeds is not a simple monotonous increase, but rather a multi-decadal fluctuation process caused by climate system changes against the backdrop of global warming.

[0066] In summary, the spatial distribution and long-term evolution trend of near-surface extreme wind speeds in China exhibit strong regional clustering and non-stationary characteristics. Particularly in high-altitude and complex terrain areas, not only are the absolute baseline levels of extreme wind speeds higher, but the long-term strengthening trend is also particularly dramatic. This spatial superposition effect further highlights the severe wind disaster risks faced by key and sensitive areas in my country under the background of climate change, and also poses new challenges to the safety of current engineering wind resistance standards.

[0067] Those skilled in the art will understand that the above description is merely a preferred embodiment of the present invention, and the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. This is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0068] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A method for reducing the time scale of near-surface annual maximum wind speed data considering spatiotemporal correction, characterized in that, The method includes the following steps: Step 1: Divide the study area into several wind speed zones based on the national basic wind speed distribution map or industry standard wind speed zoning map, and calculate the average wind speed in different seasons within each zone, taking into account the seasonal characteristics of wind speed in the zone. Step 2: For the hourly wind speed sequence within any grid point U(t) Using the average wind speed of different seasons within the area to which the grid point belongs. The mean-free processing was performed to obtain the fluctuating wind speed sequence. ; Step 3: Based on the fluctuating wind speed sequence obtained in Step 2 The Welch average periodogram method was used to estimate fluctuating wind speed. power spectral density ; Step 4: Based on power spectral density Calculate the zeroth and second moments; Step 5: Wind speed spectrum based on the original wind speed sequence Reconstructed and corrected mixed wind speed spectrum The mixed wind speed spectrum It includes a low-frequency band and a high-frequency band. The low-frequency band retains the original power spectral density pattern, while the high-frequency band adopts... Tail compensation is performed using the power law, where the starting frequency of the high-frequency band is a preset cutoff frequency and the ending frequency is the target frequency. Step 6: Based on the mixed wind speed spectrum Corrected spectral moments and correction peak factor; Step 7: Calculate the corrected annual maximum expected wind speed using the corrected spectral moments and correction peak factor; Step 8: Compare the corrected annual maximum expected wind speed with the observed annual maximum wind speed from the original sample sequence. Compare the results and output the corrected 10-minute annual maximum wind speed.

2. The method for processing annual maximum wind speed drop timescale data considering spatiotemporal correction according to claim 1, characterized in that, Step 1 also includes taking into account the spatial and seasonal effects of wind speed distribution patterns, thereby achieving the accurate pulsed wind speed de-averaging in Step 2.

3. The method for processing annual maximum wind speed drop timescale data considering spatiotemporal correction according to claim 1, characterized in that, In step 3, the Welch average periodogram method is used to estimate the fluctuating wind speed. power spectral density The method is as follows: Where T is the integral time scale, representing the average time of the turbulent vortex; The variance of the fluctuating wind speed. The cycle frequency.

4. The method for processing annual maximum wind speed drop timescale data considering spatiotemporal correction according to claim 1, characterized in that, Step 4 is based on power spectral density The method for calculating the zeroth and second moments is as follows: In the formula, The power spectrum of the Gaussian process u(t) is given by angular frequency. , For filters, For averaging time, the original spectral moments are calculated from... =0 to The integration is capped at the Nyquist frequency and cutoff at the Nyquist frequency. An ideal low-pass filter.

5. The method for processing annual maximum wind speed drop timescale data considering spatiotemporal correction according to claim 1, characterized in that, Step 5: Wind speed spectrum based on the original wind speed sequence Reconstructed and corrected mixed wind speed spectrum The method is as follows: Set cutoff frequency As the dividing point between the model spectrum and the physical spectrum, The original model spectrum is partially preserved; a target frequency is set. As the upper limit of integration, it corresponds to the Nyquist frequency on a 10-minute timescale. The segment is the high-frequency tail portion, applied Power-law tail compensation is applied to shift the spectral lines from... Extending to the Nyquist frequency corresponding to 10 minutes .

6. The method for processing annual maximum wind speed drop timescale data considering spatiotemporal correction according to claim 1, characterized in that, In step 6, based on the mixed wind speed spectrum The methods for correcting the spectral moments and the peak factor are as follows: The formula for the corrected spectral moment is: in, Corrected wind speed variance , This is the corrected variance of the derivative; For the annual maximum wind speed event, an average number of exceedances is set. , The cycle is 1 year. For wind speed to exceed The occurrence rate of peak factors that have a significant impact on the annual maximum wind speed Umax. : in, The 10-minute average annual maximum expected wind speed. This represents the average wind speed. The standard deviation of wind speed; Substituting the modified spectral moment formula into the peak factor formula yields the modified peak factor. : The formula for the corrected spectral moment is: in, Corrected wind speed variance , This is the corrected variance of the derivative.

7. The method for processing annual maximum wind speed drop timescale data considering spatiotemporal correction according to claim 1, characterized in that, The method for calculating the corrected annual maximum expected wind speed in step 7 using the corrected spectral moments and the correction peak factor is as follows: .

8. The method for processing annual maximum wind speed drop timescale data considering spatiotemporal correction according to claim 1, characterized in that, In step 8, the corrected annual maximum expected wind speed is compared with the observed annual maximum wind speed of the original sample sequence. The method for comparing and outputting the corrected 10-minute annual maximum wind speed is as follows: .

9. A computer storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1-8.

10. A computer device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the method of any one of claims 1-8.

Images