Air traffic management (ATM) attempts to assist aircraft’s approach and landing procedures with a safety-first operation. It can be challenging to evaluate aviation economics, environmental issues, and safety operations all at once while making decisions inside the terminal maneuvering area (TMA). A comprehensive arrival plan that considers weather factors and aircraft trajectory configuration is essential to increase the job efficiency of air traffic controllers and alleviate the negative environmental impact. Current state-of-the-art solutions do not fully consider unfavorable weather circumstances and unusual aircraft paths in arrival scheduling, arrival airborne congestion, and arrival time prediction. This thesis proposes a framework for the aircraft trajectory aspect to extract fe...[ Read more ]

Air traffic management (ATM) attempts to assist aircraft’s approach and landing procedures with a safety-first operation. It can be challenging to evaluate aviation economics, environmental issues, and safety operations all at once while making decisions inside the terminal maneuvering area (TMA). A comprehensive arrival plan that considers weather factors and aircraft trajectory configuration is essential to increase the job efficiency of air traffic controllers and alleviate the negative environmental impact. Current state-of-the-art solutions do not fully consider unfavorable weather circumstances and unusual aircraft paths in arrival scheduling, arrival airborne congestion, and arrival time prediction. This thesis proposes a framework for the aircraft trajectory aspect to extract features from data and predict arrival transit time, with consideration of adverse weather and non-standard flight trajectory operations. The developed procedures can reveal more air-traffic information and patterns from data, which can be used to predict arrival transit time in various situations better. The spatio-temporal pattern identification for aircraft congestion and arrival transit time within the Hong Kong International Airport TMA was presented in this thesis. To include weather factor, the hourly recorded weather radar images were also included in the analysis.

To investigate weather impact, the author proposed a scheme to quantify the weather impact on airport arrival on-time performance, by using a growth function to represent the performance deterioration with increasingly more adverse weather conditions. The model parameters and hyperparameters were derived based on actual data via a Bayesian approach. The developed scheme could also quantify the impact of dangerous weather phenomena, which were often excluded in existing aviation weather impact studies. Results exhibited the generality and versatility of the developed weather impact quantification model, which could also be used to compare the aviation weather impact at different airports. The updateability of the Bayesian approach also allows us to consider the future climate change impact on arrivals.

Lastly, the author developed a model architecture to predict arrival transit time by considering inputs pertaining to three important aspects in air traffic operations, namely aircraft, airport, and weather. The results indicated that the developed model structure could reduce the prediction errors by 5.63% when all weather conditions were considered and 8.45% under extreme weather scenarios. The arrival transit time prediction was demonstrated with several metrics to represent weather conditions. In particular, two metrics were introduced in this work that not only contained weather information, but also its interaction with airport traffic and individual flight time. These metrics were shown to yield better arrival transit time prediction accuracy than other weather metrics used in the study.