Severe convective weather is one of the important factors affecting aircraft flight. In order to reduce the economic waste of diverting aircrafts and the probability of encountering hazards, it is necessary to accurately delineate and predict the restricted areas. The first step is to extract the data on the thunderstorm points that affect aircraft flight. The initial polygon of the static flight restricted area is delineated using Graham algorithm. The change in geometry of the flight restricted area is predicted using the distance-mean method. Markov theory is introduced to predict the change of the center point position of the flight restricted area through the class state transfer matrix. The angular increment method is proposed to predict the angular change of the center point of the flight restricted area. The results show that the prediction accuracy is high and the deviation is low for radar weather data with low temporal resolution, and the prediction area can be updated in real time. The dynamic prediction is achieved on the basis of the static flight restriction zone, which makes the prediction results closer to the actual change.