Detection and Analysis of Aviation Safety Events using Historic Flight Data

In this paper, the final approach overshoot, localizer deviation, and glideslope deviation, which frequently occur in the approach phase, were detected and their occurrence was analyzed using ADS-B data of arrivals that landed on Jeju International Airport runway 25 in 2019. The detected events were analyzed based on the defined characteristics. This paper defines and analyzes aviation safety events based on their characteristics. Safety events can be determined by start time, duration, and gap between consecutive events and these characteristics subdivide the event definition into instantaneous, short continuous, and long continuous events. For complex events, they can be defined as simultaneous or sequential events. The thresholds of the event gap and duration were determined using the distributions from the event detection results and the detected events were classified into instantaneous, continuous, and complex events. This paper also visualizes the analysis results using an event tree, which provides information about the events and their propagation in chronological order. The event tree shows the process by which different events are derived from the first event. In addition, it helps to predict the subsequent events and the probability of their occurrence after the specific event has occurred.