Chaotic characteristics of traffic flow time series is analyzed to further investigate nonlinear characteristics of air traffic system.Phase space is reconstructed both by time delay which is built through mutual information,and by embedding dimension which is based on false nearest neighbors method.In order to analyze chaotic characteristics of time series,correlation dimensions and the largest Lyapunov exponents are calculated through Grassberger-Procaccia(G-P)algorithm and small-data method.Five-day radar data from the control center in Guangzhou area are analyzed and the results show that saturated correlation dimensions with self-similar structures exist in time series,and the largest Lyapunov exponents are all equal to zero and not sensitive to initial conditions.Air traffic system is affected by multiple factors,containing inherent randomness,which lead to chaos.Only grasping chaotic characteristics can air traffic be predicted and controlled accurately. Chaotic characteristics of traffic flow time series is analyzed to further investigate nonlinear characteristics of air traffic system. Phase space is reconstructed both by time delay which is built through mutual information, and by embedding dimension which is based on false nearest neighbors method.In order to analyze chaotic characteristics of time series, correlation dimensions and the largest Lyapunov exponents are calculated through Grassberger-Procaccia (GP) algorithm and small-data method. FIVE-day radar data from the control center in Guangzhou area are analyzed and the results show that saturated correlation dimensions with self-similar structures exist in time series, and the largest Lyapunov exponents are all equal to zero and not sensitive to initial conditions. Air traffic system is affected by multiple factors, containing inherent randomness, which lead to chaos.Only grasping chaotic characteristics can air traffic be predicted and controlled accurately.