In order to reduce the impact of severe weather on the normal flight route, it is necessary to plan and forecast severe weather scientifically. The improved Graham algorithm is used to determine the convex polygons formed in the forbidden area in severe weather. By establishing the gray prediction model, using the historical data of the initial boundary point of the flight forbidden area in severe weather, this study forecast the initial boundary point of the flight forbidden area caused by severe weather dynamically. Then, extending the predicted boundary points 25km outward along the central point, it tried to determine the boundary of the forbidden area of the severe weather in the future. Finally, an example about severe weather in Fuzhou was studied. The results show that the method proposed in this paper is suitable for the boundary prediction of flight forbidden area in severe weather flight in the short period of time. The accuracy is about 85.3%.