Abstract:Based on the observation data from 44 Automatic Weather Stations(AWS) in Yuyao and the reanalysis data ERA5 from ECMWF during 2007 to 2017, a statistical analysis on daily maximum wind gust of each AWS was conducted and a Localized Empirical Wind Gust Prediction Model(LEWGPM) without considering thunderstorm day was proposed. Results show that daily maximum wind gust is normally distributed with the peak of wind scale 4-5. The averaged gale day per year is 26, which account for 7.03% yearly and 3.32% hourly. It also shows the weaker the wind, the better the liner fitting between wind gust and mean winds. Some surveys on the factors that may influence the wind gust prediction were made and 10 meter mean winds, 925 hPa mean winds, roughness, friction velocity and 3 hours pressure change appear good reactions were found out. In the LEWGPM below or equal to wind scale 6, traditional empirical coefficient 1.4 has the risk of weakening the wind gust forecast and the new empirical coefficient 1.775 will lead to better forecast. When it comes to LEWGPM above wind scale 6, the mean winds, downward momentum transportation, horizontal momentum transfer, roughness and latitude are took into consideration, which found out it liner fitting to squared 925 hPa mean winds and logarithmic correlation to friction velocity. Compared to traditional empirical coefficient forecast, LEWGPM above wind scale 6 promoted 59.61%. Finally, the LEWGPM with data from 2018 to 2019 was tested, which figure out that its prediction strengthens wind scale 1-2, weakens wind scale 6, fits well with wind scale 3-5, and the accuracy of wind scale over 6 is 55.2%. The topographic friction is particularly important during typhoons and cold air events. Both the two gale processes share the similar distribution of gust factor but the different downward momentum, which means the typhoons always transfer more momentum downward.