Sensitivity And Fusion Analysis Based On WRF-LES 100-meter Wind Field Simulatio

The Wind is an integral part of refined meteorological products,and it is also one of the vital forecasting elements that the government and the public attach importance to.The instantaneous change of wind field is highly heterogeneous,so it is of great significance to grasp the instantaneous change of wind field and its spatial distribution in time and accurately for wind resource assessment,significant activity guarantee,and gale-related disaster warning.With the increasingly refined requirements of meteorological support,high-resolution and accurate wind fields are increasingly in demand.As the most potent tool for obtaining the meteorological field,numerical simulation has absolute advantages in spatiotemporal resolution and continuity.However,numerical simulation still has many uncertainties in physical parameters and the initial background field.Therefore,based on the mesoscale numerical Weather Research Forecasting（WRF）nested Large Eddy Simulation（LES）with the Global atmospheric real-time analysis product（ART-Global）developed by China Meteorological Administration（CMA）,the 1 km boundary layer scheme and the parameters of the subgrid model in LES were investigated in this study.The hectometer wind field with soil temperature and soil moisture products of the CLDAS（CMA’s Land Data System）as initial fields was simulated.The Space-Time Multiscale Analysis System（STMAS）was used to integrate the observation information.The results were evaluated and verified systematically,providing a scientific basis and reference for the applying of high-resolution wind fields.The main conclusions of this study are as follows:（1）Sensitivity experiments of 1 km boundary layer schemes and subgrid models:the sensitivity experiment results showed that the YSU boundary layer scheme had a better simulation effect than the LES boundary layer scheme at medium and micro-scale transition resolution of kilometers and SMAG（Smagorinsky）had the best performance in subgrid models.1 km used the YSU boundary layer with low deviation and root-mean-square error while maintaining the ability of capturing details of turbulence.The four subgrid models had similar performances in wind field simulation,among which SMAG had the best performance,which was manifested as the smallest root mean square error,the highest correlation coefficient,also the wind vector distribution,wind speed and direction probability density distribution and instantaneous spatial distribution of wind field were most close to the observation.There was an overestimation in the Nonlinear Backscatter and Anisotropy models（NBA）,but NBA had an advantage in capturing details regarding wind field distribution.However,the wind speed simulation of WRF-LES still underestimates the high value and overestimates the low value.（2）Sensitivity experiments of the soil temperature and moisture initial field to wind field simulation:the results showed that the ART-Global product and the land surface product of CLDAS have the reproducibility of spatial distribution of the soil temperature and soil moisture.The ART-Global product was higher according to the observation,while the land surface product of CLDAS were closer to the observation.Replacing the initial soil temperature and soil moisture field can improve wind field simulation inordinately:the mean deviation and root mean square error of the total wind speed and U component can be improved by 0.04～0.3m·s^-1 by replacing the soil moisture products of CLDAS alone.Also,1 m·s^-1 mean deviation can be improved the of the V component by replacing CLDAS soil moisture and soil temperature field.From the wind speed distribution in horizontal and vertical sections,different experiments can capture the passing by of wind.However,there were slight differences in wind speed value and spatial distribution.The ART-Global product simulated the maximum wind speed among all the experiments,while the simultaneous replacement of soil temperature and soil moisture simulated the minimum wind speed.This phenomenon may be related to the higher soil temperature and soil moisture in The ART-Global product,resulting in stronger turbulence and increasing simulated wind speed.（3）Fusion analysis of wind field simulation results:the STMAS analysis results were significantly improved.It showed that the mean deviation in the non-independent evaluation was all below 1.7 m·s^-1,and the root means square error was reduced by about 40%.The mean deviation and root mean square error of the independent evaluation of wind field fusion analysis were consistent with the distribution of the non-independent evaluation.The time series of wind speed also showed the improvement of simulated high values after STMAS analysis.The ART-Global product,replacing with the soil temperature and soil moisture in CLDAS separately were similar in the scatter distribution of the wind vector and the frequency distribution of wind speed and direction.With the integration of the observation information,the wind vector distribution of all experiments was more concentrated,and the peak shape and peak landing position of the probability distribution were also improved with varying degrees.