The real-time flight trajectory data of Jing-Jin-Shi multi-airport system was obtained by tracking and filtering， and an integrated detection framework for air flow operation structure of multi-airport system including trajectory network， trajectory cluster flow pattern， trajectory tube in crossing airspace and trajectory phase state was built. Through classification， grouping， layered and phased detection of adjacent flow， concurrent flow， multivariable flow and collective flow， it is found that： A small number of central trajectories can form a dominant air flow operation. The four groups of trajectory cluster flow patterns all last for a long time and there are differences in the concentration and dispersion of sliding windows. The spatio-temporal interaction of air traffic flow among the trajectory tube in crossing airspace is shown as a crossing relationship and also shares the same airspace unit at the same time. Besides， the phase state changes due to the influence of air flow participation and aviation network. Compared with the New York multi-airport system in the United States， the central trajectories are less and the flow direction area is narrower. Secondly， the trajectory cluster flow pattern and the trajectory tube in crossing airspace have longer concentration time and less variability， and the trajectory congestion is more likely and more frequent. The participation of Beijing Daxing International Airport will increase the central trajectories， reduce the peak value of sliding windows and also make the new crossing airspace. This research reveals the fundamental role of the aviation geographic market， which can provide decision-making for the dynamic airspace configuration and coordination of multi-airport system.