Based on the wind data at four weather stations in Zhoushan and ERAinterim data during 2008-2016, the climatic characteristics of gust factors with the distribution of mean wind speed, wind direction, hour and month are analyzed respectively. The correlation analysis is carried out between gust factors and atmospheric stability, the ratio of the wind speed at the levels of 250 to 1000 m to the wind speed of 10 m, and the 6h temperature variation at different levels in the atmospheric boundary layer. Then the best predictors are selected and the cyclic tests of gust forecast are conducted based on the BP artificial neural network method. The results show that (1) When the mean wind speed is small, the gust factors fluctuate greatly. The gust factors of air flows near the mainland or from the land direction are larger. (2) The gust factors are relatively larger in 11:00-16:00, when the turbulence affected by solar radiation is relatively strong. The gust factors from July to September and November to December are also relatively larger owing to typhoon influences or the larger surface roughness respectively. (3) The relations of the gust factors with the ratio of the wind speed of 250-1000 m to the wind speed of 10 m and temperature present positive correlation, and there presents negative correlation with stability parameters at the atmospheric boundary layer. It proves that the main physical cause of the gust is relevant to the vertical turbulent transport of momentums. (4) The cyclic tests reveal the important role of stability and turbulence at the atmospheric boundary layer in gust forecast. The gust absolute errors and their variances of the optimum groups are reduced by 10% to 25%, compared with those of comparison groups at four stations.