Research On Verification And Correction Of WRF Model Meteorological Element Simulation Accuracy

With the acceleration of China's urbanization,people have more and more cars,more and more environmental pollution and more and more serious haze phenomenon.The prediction effect of haze and other phenomena is closely related to the reliability of meteorological elements provided by mesoscale meteorological models.Because of the uncertainty and randomness of meteorological elements,it is very important to make accurate prediction.Based on the observation data of weather stations,this study adopts WRF(numerical weather prediction)model to predict meteorological elements.In order to improve the accuracy of WRF model prediction,the errors between the measured meteorological data and WRF simulated data in the southern margin of Xinjiang and Tianjin are respectively compared,and the meteorological elements output by WRF model are corrected by Geographic weighted regression (GWR).The main conclusions are as follows:(1)For wind speed,temperature and relative humidity in the southern margin of Xinjiang,the simulation results of YSU(non-local K theory)scheme are generally superior to those of MYJ(local mellor-yamadas 2.5 stage)scheme.In Tianjin,the simulation results of two boundary layer parameterization schemes are very similar.Therefore,different boundary layer parameterization schemes should be selected for meteorological element simulation of different topography.(2)WRF model can basically simulate the temperature and relative humidity characteristics of the southern margin of Xinjiang and Tianjin,but the simulation of wind speed is relatively large.The simulation accuracy of WRF model for temperature,relative humidity and wind speed in the southern margin of xinjiang is lower than that of Tianjin.The simulation error is related to the terrain.(3)The simulation results of WRF model are revised by adopting the method of geographic weighted regression.The corrected meteorological elements are closer to the observation data,and the root-mean-square error and relative root-mean-square error are significantly reduced.w ith significant correction effect.When the error of model simulation results is larger and the prediction results are less than ideal,the correction effect of geographic weighted regression is better and the error reduction range is larger after the correction.The main reason is that geographical weighted regression takes spatial heterogeneity into account and the correction process is more scientific.The simulation effect of WRF model under different terrains is related to the climate and terrain characteristics of different regions.Geographically weighted regression model in the regression parameters embedded in each station location,take into consideration of the local estimation of parameters,to determine the weight is based on local characteristics,different areas of spatial nonstationarity in space matrix,said such a spatial location relationship between the variables are enhanced adaptability,the result of using this model to simulate more accord with actual situation,make the correction results and the observation data more close,correction effectis better.