Research On Refined Wind Field Forecast In East China Based On Multi-model Integration

Based on the operational wind forecast of European Centre for Medium-Range Weather Forecasting(ECMWF),Global Forecast System(GFS)of National Centers for Environmental Prediction,Global/Regional Assimilation and Prediction System Meso(GRAPES-Meso)and Global Forecast System(GRAPES-GFS)of National Meteorological Center of China,multiple multi-model integrated forecasting method is tested and valued after statistical downscaling in East China and adjacent area.Following main conclusions are obtained:(1)In the application of temporal and spatial statistical downscaling of wind fields,the linear interpolation algorithm performs better in high-temporal-resolution interpolation and the inverse distance weighting method(IDW)performs better in high-horizontal-resolution interpolation.In addition,by introducing topographic and situ-observation,statistical revisions using multiple linear regression algorithm can effectively reduce prediction errors compared to directly interpolated forecasts for all models.(2)The integrated multi-model forecasting based on the Augmented complex extended Kalman filter(ACEKF)method can better improve the forecasting skill and stability of wind speed and direction compared bias removed ensemble averaging(BREM),linear regression-based multi-model super ensemble forecasting(SUP)and individual models.Improvements of the ACEKF for high altitude wind speed and direction forecasts are larger than that of surface wind speed and direction forecasts,(3)Using situ wind observation data from Shanghai Hongqiao Airport,Qingdao Liuting Airport and Xiamen Gaoqi Airport,a targeted comparative analysis of the effectiveness of different integration methods for forecasting low-level wind speeds near airports is conducted.Results show that bias removed averaging(BREM)and simple ensemble averaging(EMN)are vulnerable to the errors of individual members and thus perform poorly overall.However,ACEKF method can effectively avoid impact of individual members on the overall prediction skill and improve forecast reliability.