Abstract
This study utilizes observed data of maximum gusts within hours, wind speed and temperature data from ECMWF deterministic model forecasts, and ERA5 reanalysis data from January to December 2021. Based on 12 reference observation stations in the eastern region of China, a gust forecast method combining traditional statistical methods and machine learning method is established to further improve the accuracy of gust forecasts. In this gust forecasting method, the gust factor method reflects the general relationship between mean wind speed and gusts, while the machine learning method quantifies the influence of the vertical distribution of upper-level wind speeds and temperatures on surface gusts. The frequency matching method, along with a disturbance coefficient calculated using ERA5 reanalysis data, addresses the variations in gust forecasting capabilities across different stations and enhances the prediction of high magnitude gusts. Testing the gust forecasting method using data from 2022 showed that the inclusion of the upper-level gust impact model based on machine learning methods can effectively reduce underestimation in the gust factor method. The incorporation of frequency matching methods significantly reduces the negative bias in the forecast of high magnitude gusts. In cases of strong winds influenced by cold air and cyclones in 2022, the gust forecasting method developed in this study performed well for high magnitude gusts.
© 2025 American Meteorological Society. This is an Author Accepted Manuscript distributed under the terms of the default AMS reuse license. For information regarding reuse and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).
