A method, medium and program product for improving TCBL gray area scale perception similarity function based on LES simulation optimization

Abstract

The present invention discloses a method, medium and program product for improving the gray-zone scale perception similarity function of TCBL based on LES simulation optimization, aiming to improve the prediction accuracy of the numerical weather forecast model for the path and intensity of tropical cyclones. The present invention generates a high-resolution benchmark data set of TCBL of different intensities by constructing a high-resolution LES simulation framework, calculates the analytical-scale turbulent kinetic energy and sub-grid turbulent kinetic energy, and constructs an improved gray-zone scale perception similarity function based on multi-scale filtering and function fitting. The similarity function can more accurately characterize the sub-grid turbulence characteristics at gray-zone resolution, and verifies its effectiveness in TC path and intensity prediction by integration with the numerical weather forecast model. The present invention significantly improves the simulation accuracy of the model under complex wind shear and strong rotation conditions, enhances the universality of the model, provides important support for meteorological forecasting and tropical cyclone research, and has broad scientific value and application prospects.

Description

Technical Field

[0001] The present invention belongs to the technical fields of atmospheric science, numerical weather forecast and computational fluid dynamics, and relates to tropical cyclone boundary layer (TCBL) turbulence parameterization, large eddy simulation (LES) and scale-aware parameterization schemes, and specifically provides a method, medium and program product for improving TCBL gray zone gray zone scale-aware similarity function based on LES optimization, and is used to improve the TCBL wind field simulation accuracy and tropical cyclone forecast level under gray zone resolution. Background Art

[0002] In numerical weather prediction (NWP) models, boundary layer parameterization schemes are mainly used to simulate turbulent mixing processes in the atmospheric boundary layer. These schemes are based on the generation, transfer and dissipation mechanisms of turbulent kinetic energy, and calculate the transfer of momentum, heat and water vapor in the atmospheric boundary layer by turbulence. However, traditional boundary layer parameterization schemes usually assume that the scale of turbulence is much smaller than the grid scale of the model (i.e., sub-grid scale), and are therefore only applicable to coarser resolutions. At gray zone resolution, the grid scale is close to the characteristic scale of turbulence, and some turbulent structures are explicitly resolved, resulting in the inability of traditional sub-grid scale turbulence models to accurately characterize the generation and dissipation processes of turbulence.

[0003] As a key component of tropical cyclones, the tropical cyclone boundary layer (TCBL) plays a vital role in regulating the exchange of momentum, heat and water vapor between the ocean and the atmosphere. Accurately simulating the turbulent structure in the tropical cyclone boundary layer (TCBL) is of great significance for improving meteorological forecast models and improving the prediction of tropical cyclone paths and intensities. However, traditional boundary layer parameterization schemes are mainly designed for typical atmospheric conditions and perform poorly in simulating tropical cyclones. This inapplicability leads to significant defects in the model in capturing the interactions between the storm core and its environment. In addition, at the gray range resolution (0.1-1 km), the presence of small-scale structures such as vortices further exacerbates the difficulty of parameterization, resulting in increased uncertainty.

[0004] In order to solve this problem, scale-aware boundary layer parameterization schemes have been proposed in recent years to deal with the gray zone problem. Large eddy simulations (LES) are used to study the turbulence characteristics at different spatiotemporal scales, and empirical scale-dependent functions are established to distinguish the turbulent fluxes analyzed in the model from the subgrid-scale turbulent fluxes, thereby reducing the overestimation of subgrid turbulent processes at gray interval resolution. However, the existing scale-aware parameterization schemes are mainly developed based on idealized convective boundary layer (CBL) and stable boundary layer (SBL) conditions, and fail to fully consider the special dynamic characteristics of TCBL. In particular, in the TCBL environment with strong wind shear and rotation effects, under different wind speed conditions, the subgrid-scale turbulent kinetic energy (SGS-TKE) shows different scale dependencies, and this change needs to be reasonably characterized by improving the similarity function.

[0005] Large Eddy Simulation (LES) is a small-scale simulation method that can explicitly resolve large-scale turbulent eddies and parameterize sub-grid turbulence. By directly simulating the large-scale turbulent structure in the atmospheric boundary layer, LES can capture the spatiotemporal distribution characteristics of turbulence and provide high-resolution wind field and turbulence data. The data obtained through LES simulation can not only serve as a benchmark for evaluating scale-aware boundary layer parameterization schemes, but can also be used to improve sub-grid turbulent kinetic energy models and optimize the selection and adjustment strategies of similarity functions. However, current research lacks a systematic method to use LES results to improve the TCBL gray zone scale-aware parameterization scheme.

[0006] In summary, due to the special turbulence characteristics of TCBL and the limitations of the existing scale-aware boundary layer parameterization scheme, it is difficult for existing technologies to accurately simulate TCBL wind fields at gray-zone resolution. In particular, under strong rotation and wind shear conditions, the generation and dissipation models of subgrid turbulent kinetic energy have significant deviations and cannot reasonably characterize the turbulence characteristics under different wind speed conditions. Therefore, it is an urgent technical problem to develop a method based on LES simulation optimization to improve the similarity function of TCBL gray-zone scale-aware parameterization, so as to improve and optimize the similarity function, dynamically adjust the generation and dissipation model of subgrid turbulent kinetic energy (SGS-TKE), adapt to the turbulence characteristics at gray-zone resolution, and improve the prediction accuracy and reliability of tropical cyclones (TC). Summary of the invention

[0007] 1. Purpose of the invention

[0008] Aiming at the defects and shortcomings of the existing TCBL parameterization scheme, such as inaccurate characterization of sub-grid turbulent kinetic energy (SGS-TKE) generation and dissipation at gray zone resolution (0.1-1km), the existing scale-aware similarity function is mainly developed based on conventional boundary layer and has poor applicability to TCBL, and it is difficult to dynamically adjust the energy transfer between the grid scale and the characteristic scale of the boundary layer, in order to solve at least one of the above-mentioned and other technical problems in the prior art, the present invention aims to provide a method, medium and program product for optimizing and improving the gray zone scale-aware similarity function of TCBL based on LES simulation. By constructing a high-resolution TCBL idealized LES simulation framework, the turbulent characteristics of TCBL at different scale resolutions are finely simulated and analyzed, the sub-grid scale turbulent kinetic energy (TKE) ratio at different scale resolutions is obtained, and the scale-aware similarity function is optimized and improved based on the LES simulation results, so as to achieve a more accurate characterization and parameterization of the TCBL turbulent characteristics, and ultimately improve the TCBL wind field simulation accuracy and tropical cyclone TC forecast level at gray zone resolution.

[0009] (II) Technical solution

[0010] In order to achieve the purpose of the invention and solve the technical problems, the present invention adopts the following technical solutions:

[0011] The first invention object of the present invention is to provide a method for optimizing and improving the TCBL gray-zone scale perception similarity function based on LES simulation, which is used to optimize and improve the gray-zone (0.1-1 km) scale perception boundary layer parameterized similarity function through tropical cyclone boundary layer (TCBL) wind field data generated by high-resolution large eddy simulation (LES), so as to improve the accuracy of the subgrid-scale turbulent kinetic energy (SGS-TKE) model at gray-zone resolution, thereby improving the accuracy of tropical cyclone path and intensity prediction. The method comprises at least the following steps when implemented:

[0012] SS1. Construction of TCBL high-resolution LES simulation framework

[0013] An idealized TCBL based on the modified Rankine vortex model is constructed, where the horizontal range of the TCBL simulation area is set to 5-10 km, the vertical range is set to 3-5 km, the horizontal grid resolution is set to 10-100 m, the vertical grid resolution is set to half of the horizontal grid resolution below 2 km, and a graded grid of 10-15 m is used above 2 km; the gradient wind speed is specified according to the preset TC intensity and size V , Radial Gradient , Maximum wind speed radius R max and the power law exponent that controls the decay of tangential wind speed with radial distance n The tangential wind speed and radial wind speed profiles are initialized, and the potential temperature and specific humidity profiles are initialized based on the composite analysis results of TC sounding observation data; periodic boundary conditions are adopted in two horizontal directions, with a no-slip lower boundary at the bottom and a wind speed free-slip boundary condition and a constant potential temperature boundary condition at the top, and a Rayleigh damping layer is set in the area above 2 kilometers to suppress gravity wave reflection;

[0014] SS2. Generate simulated data sets with different TCBL intensities

[0015] Based on the constructed TCBL high-resolution LES simulation framework, LES simulations are carried out for multiple different wind speed conditions within the range of 20~50m / s at a height of 10 meters. The large-scale turbulent eddies in TCBL are explicitly resolved, and the SGS-TKE closed model is used to parameterize the unresolved sub-grid scale turbulence. The simulation duration is set to 6~8 hours and three-dimensional field data are output every 5~10 minutes. The simulation is ensured to reach a quasi-steady state by monitoring the evolution of the boundary layer height and turbulence statistics. A high-resolution benchmark dataset containing at least three-dimensional wind field, temperature field, humidity field and turbulence characteristics is generated based on the LES simulation results to optimize and improve the gray zone scale perception similarity function.

[0016] SS3. Constructing a similarity function for gray-area scale perception in TCBL

[0017] Based on the high-resolution LES simulation data set of TCBL with different wind speed intensities, the analytical scale turbulent kinetic energy of the TCBL wind field at high resolution is calculated. e res and subgrid turbulent kinetic energy e sgs Then, a two-dimensional Gaussian filter is used to perform multi-scale filtering on the high-resolution LES wind field and separate different filter scale resolutions in the range of 20m~2.56km. Δx Analytical wind field data under different filter scale resolutions Δx Analytical scale turbulent kinetic energy of the wind field under TCBL eres ( Δx ), and then calculate the resolution of different filter scales Δx Subgrid turbulent kinetic energy under e sgs ( Δx ) and its share of the total turbulent kinetic energy e TOT The proportion P sgs e , and based on the calculation results, fit P sgs e and Δx / h The functional relationship between h is the boundary layer height, and an improved TCBL gray area scale perception similarity function is constructed;

[0018] SS4. Validation and implementation of the improved TCBL gray-scale perceptual similarity function

[0019] The optimized and improved TCBL gray-zone scale-perceived similarity function in step SS3 is integrated into the existing numerical weather forecast model, and backtested through a series of historical data of TC events. The simulation results using the improved similarity function are compared with those using the existing similarity function. The performance differences of the models before and after the improvement in predicting TC paths and intensities are compared, and the effect of the improved similarity function on improving the TCBL wind field simulation accuracy and tropical cyclone forecast level is evaluated to ensure its applicability and accuracy at gray-zone resolution.

[0020] The second inventive object of the present invention is to provide a computer program product, including computer instructions, for executing the above-mentioned method for improving the TCBL gray zone scale perception similarity function based on LES simulation optimization.

[0021] The third inventive object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for improving the TCBL gray zone scale perception similarity function based on LES simulation optimization.

[0022] (III) Technical Effect

[0023] Compared with the prior art, the method, medium and program product of the present invention for improving the TCBL gray area scale perception similarity function based on LES simulation optimization have the following beneficial and significant technical effects:

[0024] (1) This paper generates a benchmark dataset of TCBL of different intensities through high-resolution large eddy simulation (LES), and improves the similarity function of gray-zone (0.1-1 km) scale perception based on the LES simulation results. By clearly analyzing the distribution characteristics of turbulent kinetic energy (TKE) and sub-grid turbulent kinetic energy, and fitting the functional relationship between the sub-grid turbulent kinetic energy ratio and the resolution parameter, the limitations of the existing parameterized model in dealing with gray-zone scales are effectively solved. The improved similarity function can more accurately capture the characteristics of boundary layer turbulence, especially under complex wind shear and strong rotation conditions, and its performance is significantly better than traditional methods.

[0025] (2) The present invention integrates the improved similarity function into the numerical weather forecast model and backtests historical tropical cyclone events. The present invention verifies the reliability of the improved model in path and intensity prediction. Specifically, the simulation results of boundary layer height and turbulent kinetic energy are more consistent with actual observations, and the path deviation and intensity error are significantly reduced. This effect is of great significance for improving the scientificity and timeliness of tropical cyclone forecasts, especially in reducing disastrous impacts.

[0026] (3) The present invention verifies the applicability of the improved similarity function under different environmental conditions by verifying multi-source data and analyzing the sensitivity of tropical cyclone events in different sea areas. The present invention also provides a scientific basis for the optimization of parameterization schemes. In particular, at the gray scale resolution, a similarity function model with dynamic adaptability is proposed by introducing multi-scale filtering analysis and data fitting technology of turbulence characteristic data. This method not only promotes the application of LES simulation technology in the field of meteorology, but also provides a reference for the study of complex boundary layer problems in future numerical weather forecast models. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flowchart of the implementation of the method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization provided by an embodiment of the present invention.

[0028] Figure 2 The different boundary layer conditions (V45 of TCBL, CBL and SBL) are shown in various Normalized profiles of turbulent kinetic energy under different values, where (a)-(c) are respectively the values ​​of V45 at =0.129, 0.345, 0.56, (d)-(f) are the cross sections under the CBL condition. =0.139, 0.362, 0.556, (g)-(i) are the cross sections of SBL under =0.146, 0.365, 0.584.

[0029] Figure 3 Shows that at a height of 0.4 h Schematic diagram of the turbulence disturbance field at , where (a) is the turbulence disturbance field of V45 of TCBL, (b) is the turbulence disturbance field of CBL and (c) is the turbulence disturbance field of SBL. Each column in (a) to (c) shows a different disturbance component field, and the 1st to 3rd columns correspond to the velocity disturbance components u '、 v 'and w ′, the first row shows the grid resolution Δx = 10 m, the second row shows the grid scale ratio The result is ≈0.36. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical scheme and advantages of the implementation of the present invention clearer, the technical scheme in the embodiment of the present invention will be described in more detail below in conjunction with the drawings in the embodiment of the present invention. The described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain the present invention, and should not be understood as limiting the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0031] The present invention aims to provide a method, medium and program product for optimizing and improving the gray-zone scale perception similarity function of TCBL based on LES simulation, which is used to optimize and improve the gray-zone (0.1-1 km) scale perception boundary layer parameterized similarity function of tropical cyclone boundary layer (TCBL) wind field data generated by high-resolution large eddy simulation LES, so as to improve the accuracy of subgrid-scale turbulent kinetic energy (SGS-TKE) model at gray-zone resolution, thereby improving the accuracy of tropical cyclone path and intensity prediction. As a specific example, such as Figure 1 As shown, the method of improving the TCBL gray area scale perception similarity function based on LES simulation optimization of the present invention mainly includes the following steps when implemented:

[0032] SS1. Construction of TCBL high-resolution LES simulation framework

[0033] An idealized TCBL based on the modified Rankine vortex model is constructed, where the horizontal range of the TCBL simulation area is set to 5-10 km, the vertical range is set to 3-5 km, the horizontal grid resolution is set to 10-100 m, the vertical grid resolution is set to half of the horizontal grid resolution below 2 km, and a graded grid of 10-15 m is used above 2 km; the gradient wind speed is specified according to the preset TC intensity and size V , Radial Gradient , Maximum wind speed radius R max and the power law exponent that controls the decay of tangential wind speed with radial distance n The tangential wind speed and radial wind speed profiles are initialized, and the potential temperature and specific humidity profiles are initialized based on the composite analysis results of TC sounding observation data; periodic boundary conditions are adopted in two horizontal directions, with a no-slip lower boundary at the bottom and a wind speed free slip boundary condition and a constant potential temperature boundary condition at the top, and a Rayleigh damping layer is set in the area above 2 kilometers to suppress gravity wave reflection.

[0034] As a preference, the gradient wind speed V The value range can be set between 35-65m / s and can be set specifically according to the target wind speed: for a tangential wind speed of 25m / s at a height of 10 meters, V =44m / s, 35m / s V =58m / s, 45m / s V =65m / s; through the power law relationship Calculate the radial gradient of tangential wind speed , where 25m / s and 35m / s are taken n =0.7, 45m / s n =0.75; Maximum wind speed radius R max The selection corresponds to different target wind speeds: 25m / s R max =110km, 35m / s R max =75km, 45m / s R max =40km.

[0035] As a preferred method, the computational domain and grid can be set as follows: the horizontal range is 5km×5km, the vertical height is 3km; the horizontal direction uses a uniform grid with a uniform resolution of 10m, and the vertical direction uses a graded grid with a resolution below 2km. Δz =5m, 2-3km interval Δz=12.5m; the thickness of the Rayleigh damping layer is set to 1km, the damping coefficient increases linearly with height, and the maximum damping coefficient is 0.003s -1 ; A buffer zone of 100 grid points is set around the computational domain to reduce numerical instability at the boundary.

[0036] As a preferred method, when initializing the thermodynamic profile based on TC sounding observation data, a quadratic function is used at the bottom 200 m of the boundary layer. Fitting potential temperature profile, The logarithmic function of is fitted to the specific humidity profile, where z is the height, θ 0 and q 0 and 0 are the initial potential temperature and specific humidity of the near-surface layer, z 0 is the feature height, a 1. a 2. b 1 is the fitting coefficient; the sea surface temperature is set to be 2°C higher than the surface air temperature, and the thermodynamic profile is smoothed in the transition zone with a thickness of 500 m at the top of the boundary layer to avoid sharp vertical gradients.

[0037] SS2. Generate simulated data sets with different TCBL intensities

[0038] Based on the constructed TCBL high-resolution LES simulation framework, LES simulations are carried out for multiple different wind speed conditions within the range of 20~50m / s at a height of 10 meters. The large-scale turbulent eddies in TCBL are explicitly analyzed, and the SGS-TKE closed model is used to parameterize the unresolved sub-grid scale turbulence. The simulation duration is set to 6~8 hours and three-dimensional field data are output every 5~10 minutes. The simulation reaches a quasi-steady state by monitoring the evolution of the boundary layer height and turbulence statistics. According to the LES simulation results, a high-resolution benchmark dataset is generated for optimizing and improving the gray-zone scale perception similarity function, which includes at least three-dimensional wind field, temperature field, humidity field and turbulence characteristics.

[0039] Preferably, in this step, the following settings are adopted for LES simulation under different wind speed conditions: the simulation duration is set to 6 hours, of which the first 4 hours are the adjustment period and the last 2 hours are the analysis period; the three-dimensional field data are output every 6 minutes during the analysis period; the convection term adopts the fifth-order upwind difference format, the time advancement adopts the third-order Runge-Kutta format, and the time step is set to 0.5 seconds; at each output moment, the turbulence statistics including the components of Reynolds stress, the vertical distribution of turbulent kinetic energy, the momentum flux profile, the effective eddy viscosity coefficient, the turbulence length scale and the potential temperature flux are calculated and recorded, and the statistically representative turbulence structure characteristics are obtained by horizontally averaging and time averaging these characteristic quantities.

[0040] Preferably, the SGS-TKE closed model in the embodiment of the present invention adopts a hybrid scheme comprising two parts: a standard Deardorff scheme for the entire computational domain and a Sullivan correction term specifically for the near-surface layer. The expression of the Sullivan correction term is: ,in τ ij is the subgrid stress tensor, C s is the dynamically calculated Smagorinsky coefficient, Δ is the filtering scale, l s is the characteristic length of the near-surface layer, | S | cl is the classical definition of shear rate, S ij is the strain rate tensor, and the Sullivan correction term only works in the area below 100 m from the ground. The simulation accuracy of near-ground turbulence is improved by considering the influence of the vertical gradient.

[0041] As a preferred method, the quasi-steady state judgment criteria are: the relative change rate of the boundary layer height and the 10-meter wind speed is less than 1% within 1 hour, and the shape of the momentum flux profile remains stable; the boundary layer height is determined by the height where the momentum flux drops to 5% of the surface value, and a large-scale inclination correction term is introduced. Maintaining the thermodynamic profile structure, where Representative temperature θ or specific humidity q , τ n The relaxation time scale is 5 minutes. The reference profile value is used; at the same time, the evolution of the sub-grid turbulent kinetic energy is monitored to ensure that its fluctuation does not exceed 5% during the analysis period.

[0042] SS3. Constructing a similarity function for gray-area scale perception in TCBL

[0043] Based on the high-resolution LES simulation data set of TCBL with different wind speed intensities, the analytical scale turbulent kinetic energy of the TCBL wind field at high resolution is calculated. e res and subgrid turbulent kinetic energy e sgs Then, a two-dimensional Gaussian filter is used to perform multi-scale filtering on the high-resolution LES wind field and separate different filter scale resolutions in the range of 20m~2.56km. Δx Analytical wind field data under different filter scale resolutions Δx Analytical scale turbulent kinetic energy of the wind field under TCBL e res (Δx ), and then calculate the resolution of different filter scales Δx Subgrid turbulent kinetic energy under e sgs ( Δx ) and its share of the total turbulent kinetic energy e TOT The proportion P sgs e , and based on the calculation results, fit P sgs e and Δx / h The functional relationship between h is the boundary layer height, and an improved TCBL gray area scale perception similarity function is constructed.

[0044] Preferably, constructing a TCBL gray zone scale perception similarity function based on a TCBL high-resolution benchmark dataset of different wind speed intensities generated by LES simulation includes at least the following sub-steps:

[0045] SS31. Computing analytic turbulent kinetic energy on high-resolution grids

[0046] Based on the TCBL benchmark dataset generated by LES simulation, the analytical scale turbulent kinetic energy of the high-resolution TCBL wind field is calculated according to the following expressions: e res and subgrid turbulent kinetic energy e sgs :

[0047]

[0048] in, u , v , w are the velocity components in three directions, represents the regional average of the wind speed component, C sgs is the empirical coefficient of the subgrid model, Δ is the high-resolution grid scale, |S| represents the shear strain rate, and its mathematical expression is: ;

[0049] SS32. Extracting analytic turbulent kinetic energy at different grid scales

[0050] A two-dimensional Gaussian filter is used to perform multi-scale filtering on the high-resolution LES wind field, separating multiple filter scale resolutions in the range of 20m~2.56km. Δx Analyze wind field data and calculate different filter scale resolutions Δx Analytical turbulent kinetic energy undere res ( Δx )as follows:

[0051]

[0052] in, For the velocity components u , v , w The components obtained after applying a two-dimensional Gaussian filter, the filtering range is determined by the grid scale Δx definition, represents the regional average of the wind speed component;

[0053] SS33. Computation of Sub-grid Turbulent Kinetic Energy e sgs ( Δx )

[0054] Based on the total turbulent kinetic energy of e TOT and the analytical turbulent kinetic energy at different grid resolutions e res ( Δx ), calculate the resolution of different filter scales Δx Subgrid turbulent kinetic energy under e sgs ( Δx ):

[0055]

[0056] in, e TOT is the total turbulent kinetic energy at high resolution and , e res , e res ( Δx ) are high resolution and filter scale resolution respectively Δx Analytical turbulent kinetic energy under ;

[0057] SS34. Calculate the proportion of subgrid turbulent kinetic energy P sgs e

[0058] Calculate the resolution of different filter scales at a given ratio of grid size to boundary layer height. Δx Subgrid turbulent kinetic energy under e sgs ( Δx ) P sgs e , whose expression is:

[0059]

[0060] in, is the total turbulent kinetic energy, , h is the boundary layer height;

[0061] SS35. Fitting improved similarity function

[0062] According to the LES simulation results under different wind speed conditions, the sub-grid turbulent kinetic energy ratio is fitted by the least squares method. P sgs e and The functional relationship between them is used to construct an improved TCBL gray area scale perception similarity function , whose expression is:

[0063]

[0064] in, a 1. a 2. a 3 is the empirical coefficient and is determined by data fitting. The improved gray area scale perception similarity function It can dynamically reflect the variation of sub-grid turbulent kinetic energy at different grid scales and adapt to the turbulent characteristics of TCBLs of different intensities.

[0065] Further preferably, the filtering operation of the above two-dimensional Gaussian filter includes the following filtering scales: Δx = 20m, 40m, 80m, 160m, 320m, 640m, 1.28km and 2.56km. For each filter scale, the analytical scale turbulent kinetic energy is calculated separately. e res ( Δx ) and sub-grid scale turbulent kinetic energy e sgs ( Δx ), and ensure that 0.05 < = Δx / h < 0.85 is satisfied. When calculating turbulent kinetic energy, the boundary effect can be handled in the following way: in the horizontal direction, the data within 100 grid points from the boundary of the computational domain are discarded, and in the vertical direction, only the data within the range of 15% to 85% of the boundary layer height are considered to ensure that the obtained turbulent statistical characteristics are representative.

[0066] In addition, in the process of fitting the TCBL gray zone scale perception similarity function, the boundary layer height is determined based on the potential temperature gradient method, the Richardson number method or the turbulent kinetic energy method. h , where: When the boundary layer height is determined based on the potential temperature gradient method h When the potential temperature θ The vertical gradient When the altitude exceeds or equals 3 K / km for the first time, the altitude is defined as the boundary layer altitude. When the boundary layer altitude is determined based on the Richardson number method, h When , first calculate the gradient of the vertical profile Ri The mathematical expression of the number is in g is the acceleration due to gravity, θ 0 is the reference potential temperature, is the vertical gradient of the horizontal average wind speed, when Ri The position where the number reaches the critical value of 0.25 is considered to have turbulence suppression, and this height is defined as the boundary layer height; when the boundary layer height is determined based on the turbulent kinetic energy method h When the total turbulent kinetic energy is defined e TOT At height h When it reaches 5% of its maximum value, this height is the boundary layer height, and its mathematical expression is ; Determine the boundary layer height h Finally, the turbulence statistics of multiple consecutive time steps are averaged to eliminate the influence of instantaneous fluctuations and ensure that the boundary layer height h accuracy and stability.

[0067] Further preferably, in the above sub-step SS35, when calculating the sub-grid turbulent kinetic energy ratio P sgse and When the functional relationship between them is established, the selection and verification process of fitting parameters can be further optimized. Similarity functions are constructed for different wind speed intensity conditions (such as 25m / s, 35m / s and 45m / s), and the fitting process and verification accuracy are optimized through the following steps:

[0068] SS35.1. Data Preprocessing

[0069] For each filter scale Δ x , remove non-physical data caused by boundary effects. In the horizontal direction, discard the data within 100 grid points from the boundary of the computational domain, and in the vertical direction, only retain the data within the range of 15% to 85% of the boundary layer height ℎ to ensure that the fitting data sample is representative.

[0070] SS35.2. Fitting under multiple wind speed conditions

[0071] Based on different wind speed intensities (such as 25m / s, 35m / s, 45m / s), the proportion of sub-grid turbulent kinetic energy is calculated. P sgse and The fitting function still takes the following form:

[0072]

[0073] in, a 1. a 2. a 3 is the empirical coefficient, which is obtained by nonlinear least squares fitting. At the same time, in order to reduce the risk of overfitting, the cross-validation method is used to evaluate the generalization ability of the fitting function to ensure that it is applicable to different wind speed intensities and boundary layer types.

[0074] SS35.3. Fitting accuracy verification

[0075] The fitting accuracy of the function is evaluated by calculating the root mean square error (RMSE) and correlation coefficient R between the fitting function and the LES simulation results. Preferably, the RMSE should be less than 0.015, and the correlation coefficient R should not be less than 0.98. The Taylor plot is further used to comprehensively analyze the performance of the fitting function under different conditions to ensure that it is within 0.05 ≤ The reliability was within the range of ≤ 0.85.

[0076] SS4. Validation and implementation of the improved TCBL gray-scale perceptual similarity function

[0077] The optimized and improved TCBL gray-zone scale-perceived similarity function in step SS3 is integrated into the existing numerical weather forecast model, and backtested through a series of historical data of TC events. The simulation results using the improved similarity function are compared with those using the existing similarity function. The performance differences of the models before and after the improvement in predicting TC paths and intensities are compared, and the effect of the improved similarity function on improving the TCBL wind field simulation accuracy and tropical cyclone forecast level is evaluated to ensure its applicability and accuracy at gray-zone resolution.

[0078] As a preferred method, the validation and evaluation process can measure the improvement effect through the following quantitative indicators: the relative root mean square error RMSE and correlation coefficient R between the SGS-TKE predicted by the similarity function and the true SGS-TKE are calculated based on the LES results, and the performance of the similarity function before and after improvement is comprehensively analyzed using the Taylor diagram; at the same time, the probability density function PDF comparative analysis method is introduced to evaluate the ability of the improved similarity function to characterize the turbulence characteristics under different TCBL heights and wind speeds, and ensure that the similarity function is within 0.05 ≤ The relative error in the range of ≤ 0.85 does not exceed 15%.

[0079] In addition, when the improved similarity function is applied to the gray-scale perception boundary layer parameterization scheme, a piecewise function form can be used to handle different resolution intervals: Δx / h The original SGS-TKE model remains unchanged in the high-resolution range <0.05. Δx / h The newly constructed similarity function is used in the gray area of ​​≤0.85. Δx / h The traditional overall parameterization scheme is used in the low-resolution range >0.85; this segmentation process ensures the continuity and physical rationality of the model at different resolutions, and cubic spline interpolation is used at the segmentation points to ensure smooth transition of the function.

[0080] At the same time, the impact of the improved similarity function on the TCBL structure can be evaluated in the following ways: calculating the turbulent kinetic energy spectrum at different heights and analyzing the distribution of energy at different scales; evaluating the vertical profile of the momentum flux and verifying the rationality of turbulent mixing in the boundary layer; calculating the ratio of turbulence intensity in the horizontal and vertical directions to test the ability of the improved scheme to characterize turbulent anisotropy; and examining the changing laws of these characteristic quantities under different wind speed conditions (such as 25m / s, 35m / s and 45m / s, etc.) to ensure the universality of the improved scheme.

[0081] In order to further verify the proportion of sub-grid turbulent kinetic energy in the gray area P sgs,e The vertical distribution characteristics under different boundary layer types are analyzed. The impact on subgrid, resolution scale and total turbulent kinetic energy, Figure 2 The results show that the TCBL has a V45, CBL and SBL boundary layer, and the Normalized profile of turbulent kinetic energy under the value of . Tropical cyclone boundary layer (V45, Figure 2 ac in the figure shows that the proportion of sub-grid turbulence increases significantly, especially in the larger The analytic turbulent kinetic energy approaches zero at low values, indicating that this type of boundary layer is more dependent on high resolution. Figure 2 The subgrid turbulence ratio is relatively low, the total turbulent kinetic energy is more evenly distributed, and the analytical turbulent energy maintains a high contribution in most height ranges. Figure 2 The proportion of turbulent kinetic energy in the gi) subgrid is higher than that in CBL, but lower than that in TCBL, and the contribution of the near-surface layer is more significant. Figure 2The results further verify the applicability and advantages of the improved similarity function of the present invention: in TCBL and SBL, the rapid increase in the proportion of sub-grid turbulent kinetic energy indicates that the existing models are difficult to characterize such boundary layer turbulence, and the improved similarity function can dynamically adapt to this characteristic; in CBL, the improved similarity function still maintains high accuracy within the gray area, further proving its universality and physical rationality. Therefore, the improved similarity function proposed in the present invention is particularly suitable for tropical cyclone boundary layers with strong turbulence characteristics and complex distribution laws, and at the same time shows good robustness and accuracy in other boundary layer types.

[0082] In this example, in order to further clarify the influence of different grid resolutions on turbulence characteristics, Figure 3 Shown is a height of 0.4 h For the three cases of TCBL V45, CBL (convective boundary layer) and SBL (stable boundary layer), the velocity disturbance component u '、 v '、 w The first row shows the grid resolution. Δx =10 m, the second line is the grid scale ratio ≈0.36. Figure 3 As shown, in the case of TCBL V45, u '、 v The two horizontal velocity disturbance components show a strip-like structure (“vortex strips”) caused by wind shear, while the vertical velocity disturbance w ′ shows eddy characteristics with a horizontal scale of about hundreds of meters, indicating that turbulence expands more significantly in the horizontal direction than in the vertical direction. When ≈0.36, the horizontal disturbance in the TCBL case retains about 27% of the initial intensity, while the vertical disturbance only retains 12%, reflecting a clear horizontal dominant characteristic. For the SBL case, its turbulent disturbance is weaker overall, but still exhibits structural characteristics similar to TCBL. After two-dimensional Gaussian filtering, the horizontal disturbance components in the SBL retain about 36% and 28%, while the vertical disturbance only retains 12%, indicating that despite its weaker intensity, the turbulence extension in the SBL also has a clear horizontal advantage. In the CBL case, u '、 v The disturbance field of ′ does not show an obvious "vortex strip" structure, but rather presents a block distribution. w The disturbance field of ′ presents an open structure of large-scale convective vortices, with horizontal scales reaching thousands of meters. ≈0.36, in the CBL case wThe interference retention rate reaches 50%, indicating that its resolution is significantly better than that of TCBL and SBL. Overall, the scale distribution of the turbulent structure of TCBL in the horizontal and vertical directions is significantly specific and is significantly affected by the grid resolution. These features further prove that higher resolution is required to accurately capture the turbulent characteristics when simulating TCBL, and the generated high-resolution dataset will provide strong support for the subsequent improvement of the similarity function.

[0083] In addition, for step SS3, in order to construct a more accurate TCBL gray area scale perception similarity function, based on Figure 3 The turbulence disturbance characteristics analysis shown in the figure, the turbulence component characteristics of different boundary layers (TCBL, CBL and SBL) at the gray scale are as follows: For the TCBL characteristics, in the V45 case, the horizontal velocity disturbance ( u '、 v ′) is manifested as a strip-shaped vortex structure dominated by wind shear, and the vertical velocity disturbance ( w ′) appear as vortices with horizontal scales of about 100 meters. These structures are ≈0.36, a certain proportion of horizontal disturbance intensity is still retained, and affected by tropical cyclone dynamic factors such as wind shear and rotation effect, the turbulence characteristics are significantly different from those of CBL and SBL. u '、 v ′The horizontal disturbance component is evenly distributed, and the vertical velocity disturbance w ′ presents the blocky characteristics of an open convective vortex, and maintains about 50% of the analytical intensity at the gray scale, indicating that its turbulent characteristics are dominated by convection, which is significantly different from the wind shear dominance of TCBL. For the SBL characteristics, the horizontal velocity perturbation structure is similar to that of TCBL, but the overall perturbation intensity is lower. The horizontal perturbation retention rate is about 36%~28% at the gray scale, while the vertical direction is only 12%, showing the significant characteristics of wind shear driving. Based on the above characteristics, when constructing the similarity function, it is necessary to comprehensively consider the analytical intensity ratio, turbulence characteristics differences and boundary layer dynamics parameters at the gray scale to optimize P sgse and The fitting accuracy between them is improved, thereby enhancing the applicability and accuracy of the gray zone scale parameterization scheme.

[0084] Through the above embodiments, the purpose of the present invention is fully and effectively achieved. Those skilled in the art can understand that the present invention includes but is not limited to the contents described in the drawings and the above specific embodiments. Although the present invention has been described with respect to the most practical and preferred embodiments currently considered, it should be understood that the present invention is not limited to the disclosed embodiments, and any modification that does not deviate from the functional and structural principles of the present invention will be included in the scope of the claims.

Claims

1. A method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization, characterized in that: The method comprises at least the following steps when implemented: SS1. Construction of TCBL high-resolution LES simulation framework An idealized TCBL based on the modified Rankine vortex model is constructed, where the horizontal range of the TCBL simulation area is set to 5-10 km, the vertical range is set to 3-5 km, the horizontal grid resolution is set to 10-100 m, the vertical grid resolution is set to half of the horizontal grid resolution below 2 km, and a graded grid of 10-15 m is used above 2 km; the gradient wind speed is specified according to the preset TC intensity and size V , Radial Gradient , Maximum wind speed radius R max and the power law exponent that controls the decay of tangential wind speed with radial distance n The tangential wind speed and radial wind speed profiles are initialized, and the potential temperature and specific humidity profiles are initialized based on the composite analysis results of TC sounding observation data; periodic boundary conditions are adopted in two horizontal directions, with a no-slip lower boundary at the bottom and a wind speed free-slip boundary condition and a constant potential temperature boundary condition at the top, and a Rayleigh damping layer is set in the area above 2 kilometers to suppress gravity wave reflection; SS2. Generate simulated data sets with different TCBL intensities Based on the constructed TCBL high-resolution LES simulation framework, LES simulations are carried out for multiple different wind speed conditions within the range of 20~50m / s at a height of 10 meters. The large-scale turbulent eddies in TCBL are explicitly resolved, and the SGS-TKE closed model is used to parameterize the unresolved sub-grid scale turbulence. The simulation duration is set to 6~8 hours and three-dimensional field data are output every 5~10 minutes. The simulation is ensured to reach a quasi-steady state by monitoring the evolution of the boundary layer height and turbulence statistics. A high-resolution benchmark dataset containing at least three-dimensional wind field, temperature field, humidity field and turbulence characteristics is generated based on the LES simulation results to optimize and improve the gray zone scale perception similarity function. SS3. Constructing a similarity function for gray-area scale perception in TCBL Based on the high-resolution LES simulation data set of TCBL with different wind speed intensities, the analytical scale turbulent kinetic energy of the TCBL wind field at high resolution is calculated. e res and subgrid turbulent kinetic energy e sgs Then, a two-dimensional Gaussian filter is used to perform multi-scale filtering on the high-resolution LES wind field and separate different filter scale resolutions in the range of 20m~2.56km. Δx Analytical wind field data under different filter scale resolutions Δx Analytical scale turbulent kinetic energy of the wind field under TCBL e res ( Δx ), and then calculate the resolution of different filter scales Δx Subgrid turbulent kinetic energy under e sgs ( Δx ) and its share of the total turbulent kinetic energy e TOT The proportion P sgs e , and based on the calculation results, fit P sgs e and Δx / h The functional relationship between h is the boundary layer height, and an improved TCBL gray area scale perception similarity function is constructed; SS4. Validation and implementation of the improved TCBL gray-scale perceptual similarity function The optimized and improved TCBL gray-zone scale-perceived similarity function in step SS3 is integrated into the existing numerical weather forecast model, and backtested through a series of historical data of TC events. The simulation results using the improved similarity function are compared with those using the existing similarity function. The performance differences of the models before and after the improvement in predicting TC paths and intensities are compared, and the effect of the improved similarity function on improving the TCBL wind field simulation accuracy and tropical cyclone forecast level is evaluated to ensure its applicability and accuracy at gray-zone resolution.

2. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS1, the gradient wind speed V The value range is 35-65m / s and is set according to the target wind speed: 25m / s at a height of 10 meters. V =44m / s, 35m / s V =58m / s, 45m / s V =65m / s; Through the power law relationship Calculate the radial gradient of tangential wind speed , where 25m / s and 35m / s are taken n =0.7, 45m / s n =0.75; Maximum wind speed radius R max The selection corresponds to different target wind speeds: 25m / s R max =110km, 35m / s R max =75km, 45m / s R max =40km.

3. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS1, the calculation domain and grid settings are as follows: the horizontal range is 5km×5km, the vertical height is 3km; the horizontal direction uses a uniform grid with a uniform resolution of 10m, and the vertical direction uses a graded grid with a resolution below 2km. Δz =5m, 2-3km interval Δz =12.5m; the thickness of the Rayleigh damping layer is set to 1km, the damping coefficient increases linearly with height, and the maximum damping coefficient is 0.003s -1 ; A buffer zone of 100 grid points is set around the computational domain to reduce numerical instability at the boundary.

4. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS1, when initializing the thermodynamic profile based on TC sounding observation data, the bottom 200m of the boundary layer is used. The quadratic function fitting potential temperature profile is The logarithmic function of is fitted to the specific humidity profile, where z is the height, θ 0 and q 0 and 0 are the initial potential temperature and specific humidity of the near-surface layer, z 0 is the feature height, a 1. a 2. b 1 is the fitting coefficient; the sea surface temperature is set to be 2°C higher than the surface air temperature, and the thermodynamic profile is smoothed in the transition zone with a thickness of 500 m at the top of the boundary layer to avoid sharp vertical gradients.

5. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS2, the LES simulation settings for different wind speed conditions are as follows: the simulation duration is set to 6 hours, of which the first 4 hours are the adjustment period and the last 2 hours are the analysis period; the three-dimensional field data are output every 6 minutes during the analysis period; the convection term adopts the fifth-order upwind difference format, the time advancement adopts the third-order Runge-Kutta format, and the time step is set to 0.5 seconds; at each output moment, the turbulence statistics including the components of Reynolds stress, the vertical distribution of turbulent kinetic energy, the momentum flux profile, the effective eddy viscosity coefficient, the turbulence length scale and the potential temperature flux are calculated and recorded, and the characteristic quantities are horizontally averaged and time-averaged to obtain statistically representative turbulence structure characteristics.

6. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS2, the SGS-TKE closed model adopts a hybrid scheme consisting of two parts: the standard Deardorff scheme for the entire computational domain and the Sullivan correction term specifically for the near-surface layer. The expression of the Sullivan correction term is ,in τ ij is the subgrid stress tensor, C s is the dynamically calculated Smagorinsky coefficient, Δ is the filtering scale, l s is the characteristic length of the near-surface layer, | S | cl is the classical definition of shear rate, S ij is the strain rate tensor, and the Sullivan correction term only works in the area below 100 m from the ground. The simulation accuracy of near-ground turbulence is improved by considering the influence of the vertical gradient.

7. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS2, the quasi-steady state judgment criteria are: the relative change rate of the boundary layer height and the 10-meter wind speed is less than 1% for 1 hour, and the shape of the momentum flux profile remains stable; the boundary layer height is determined by the height where the momentum flux drops to 5% of the surface value, and a large-scale inclination correction term is introduced Maintaining the thermodynamic profile structure, where Representative temperature θ or specific humidity q , τ n The relaxation time scale is 5 minutes. The reference profile value is used; at the same time, the evolution of the sub-grid turbulent kinetic energy is monitored to ensure that its fluctuation does not exceed 5% during the analysis period.

8. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS3, a TCBL gray zone scale perception similarity function is constructed based on the TCBL high-resolution benchmark dataset of different wind speed intensities generated by LES simulation, which includes at least the following sub-steps: SS31. Computing analytic turbulent kinetic energy on high-resolution grids Based on the TCBL benchmark dataset generated by LES simulation, the analytical scale turbulent kinetic energy of the high-resolution TCBL wind field is calculated according to the following expressions: e res and subgrid turbulent kinetic energy e sgs : in, u , v , w are the velocity components in three directions, represents the regional average of the wind speed component, C sgs is the empirical coefficient of the subgrid model, Δ is the high-resolution grid scale, |S| represents the shear strain rate, and its mathematical expression is: ; SS32. Extracting analytic turbulent kinetic energy at different grid scales A two-dimensional Gaussian filter is used to perform multi-scale filtering on the high-resolution LES wind field, separating multiple filter scale resolutions in the range of 20m~2.56km. Δx Analyze wind field data and calculate different filter scale resolutions Δx Analytical turbulent kinetic energy under e res ( Δx )as follows: in, For the velocity components u , v , w The components obtained after applying a two-dimensional Gaussian filter, the filtering range is determined by the grid scale Δx definition, represents the regional average of the wind speed component; SS33. Computation of Sub-grid Turbulent Kinetic Energy e sgs ( Δx ) Based on the total turbulent kinetic energy of e TOT and the analytical turbulent kinetic energy at different grid resolutions e res ( Δx ), calculate the resolution of different filter scales Δx Subgrid turbulent kinetic energy under e sgs ( Δx ): in, e TOT is the total turbulent kinetic energy at high resolution and , e res , e res ( Δx ) are high resolution and filter scale resolution respectively Δx Analytical turbulent kinetic energy under ; SS34. Calculate the proportion of subgrid turbulent kinetic energy P sgs e Calculate the resolution of different filter scales at a given ratio of grid size to boundary layer height. Δx Subgrid turbulent kinetic energy under e sgs ( Δx ) P sgs e , whose expression is: in, is the total turbulent kinetic energy, , h is the boundary layer height; SS35. Fitting improved similarity function According to the LES simulation results under different wind speed conditions, the sub-grid turbulent kinetic energy ratio is fitted by the least squares method. P sgs e and The functional relationship between them is used to construct an improved TCBL gray area scale perception similarity function , whose expression is: in, a 1. a 2. a 3 is the empirical coefficient and is determined by data fitting. The improved gray area scale perception similarity function It can dynamically reflect the variation of sub-grid turbulent kinetic energy at different grid scales and adapt to the turbulent characteristics of TCBLs of different intensities.

9. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 8 is characterized in that: In step SS3, the filtering operation of the two-dimensional Gaussian filter includes the following filtering scales: Δx = 20m, 40m, 80m, 160m, 320m, 640m, 1.28km and 2.56km. For each filter scale, the analytical scale turbulent kinetic energy is calculated separately. e res ( Δx ) and sub-grid scale turbulent kinetic energy e sgs ( Δx ), and ensure The conditions are met.

10. The method for improving the TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 8 is characterized in that: In step SS3, when calculating the turbulent kinetic energy, the boundary effect is handled in the following way: in the horizontal direction, the data within 100 grid points from the boundary of the calculation domain are discarded, and in the vertical direction, only the data within the range of 15% to 85% of the boundary layer height are considered to ensure that the obtained turbulence statistical characteristics are representative.

11. The method for improving TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 8 is characterized in that: In step SS3, in the process of fitting the TCBL gray area scale perception similarity function, the boundary layer height is determined based on the potential temperature gradient method, the Richardson number method or the turbulent kinetic energy method. h , where: When the boundary layer height is determined based on the potential temperature gradient method h When the potential temperature θ The vertical gradient When the altitude exceeds or equals 3 K / km for the first time, the altitude is defined as the boundary layer altitude. When the boundary layer altitude is determined based on the Richardson number method, h When , first calculate the gradient of the vertical profile Ri The mathematical expression of the number is ,in g is the acceleration due to gravity, θ 0 is the reference potential temperature, is the vertical gradient of the horizontal average wind speed, when Ri The position where the number reaches the critical value of 0.25 is considered to have turbulence suppression, and this height is defined as the boundary layer height; when the boundary layer height is determined based on the turbulent kinetic energy method h When the total turbulent kinetic energy is defined e TOT At height h When it reaches 5% of its maximum value, this height is the boundary layer height, and its mathematical expression is ; Determine the boundary layer height h Finally, the turbulence statistics of multiple consecutive time steps are averaged to eliminate the influence of instantaneous fluctuations and ensure that the boundary layer height h accuracy and stability.

12. The method for improving TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS4, the verification and evaluation process measures the improvement effect through the following quantitative indicators: the relative root mean square error RMSE and correlation coefficient R between the SGS-TKE predicted by the similarity function and the true SGS-TKE are calculated based on the LES results, and the performance of the similarity function before and after improvement is comprehensively analyzed using the Taylor diagram; at the same time, the probability density function PDF comparative analysis method is introduced to evaluate the ability of the improved similarity function to characterize the turbulence characteristics under different TCBL heights and wind speeds, and ensure that the similarity function is within 0.05< Δx / h The relative error in the range of <0.85 does not exceed 15%.

13. The method for improving TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS4, when the improved similarity function is applied to the gray-area scale-aware boundary layer parameterization scheme, a piecewise function form is used to process different resolution intervals: Δx / h The original SGS-TKE model remains unchanged in the high-resolution range <0.

05. Δx / h The newly constructed similarity function is used in the gray area of ​​≤0.

85. Δx / h The traditional overall parameterization scheme is used in the low-resolution range >0.85; the continuity and physical rationality of the model at different resolutions are ensured through segmentation processing, and cubic spline interpolation is used at the segmentation points to ensure the smooth transition of the function.

14. The method for improving TCBL gray area scale perception similarity function based on LES simulation optimization according to claim 1 is characterized in that: In step SS4, when evaluating the impact of the improved similarity function on the TCBL structure: calculate the turbulent kinetic energy spectrum at different heights and analyze the distribution of energy at different scales; evaluate the vertical profile of the momentum flux to verify the rationality of turbulent mixing in the boundary layer; calculate the ratio of turbulence intensity in the horizontal and vertical directions to test the improved scheme's ability to characterize turbulent anisotropy; examine the change law of characteristic quantities under different wind speed conditions to ensure the universality of the improved scheme.

15. A computer program product, comprising computer instructions, for executing the method for improving the TCBL gray zone scale perception similarity function based on LES simulation optimization as described in any one of claims 1 to 14.

16. A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for improving the TCBL gray zone scale perception similarity function based on LES simulation optimization as described in any one of claims 1 to 14.

Images