With the long-term goal of supporting the training of air traffic controllers and ensuring safety through real-time monitoring of operations, we propose a clustering-based framework that seeks to extract, model, and characterize the control actions of air traffic controllers when managing aircraft within the immediate vicinity of an airport. Toward achieving our goal, we begin by developing and applying a hidden Markov model to identify and extract heading changes within aircraft trajectories. Identification of these heading changes allows for translating spatially defined trajectories into variable-length strings describing executed lateral maneuvers resulting from the control actions of air traffic controllers. The trajectory strings are then compared and clustered using edit-distance metrics in combination with the -medoids clustering algorithm. Taken together, each step forms a consistent framework for modeling and understanding trajectories in the context of control actions of air traffic controllers. The proposed modeling and clustering framework also enables identification of anomalous and potentially disruptive decisions when managing aircraft. Through application on a set of historical trajectories at Washington National Airport, we are able to demonstrate that the proposed framework is able to successfully overcome shortfalls associated with traditional clustering techniques.