Development of methods for statistical modeling of air traffic demonstrated through a Winnipeg-area case study

Abstract

The integration of remotely piloted aircraft systems (RPAS) into shared airspace requires a thorough risk analysis. Specific operations risk assessment (SORA) is a widely adopted approach by international civil aviation authorities to guide RPAS operators in evaluating risks associated with their mission. A critical step in the SORA process is analyzing the airspace where the operation will take place, which requires knowledge of the intruder aircraft's flight characteristics as well as the airspace model. This paper proposes a methodology for developing a statistical airspace model using historical aircraft track data collected in the Winnipeg Manitoba Flight Information Region. The developed methods include data cleaning routines, Kalman filters for track smoothing, and Bayesian networks for synthetic track generation, following an approach similar to that employed by the Massachusetts Institute of Technology Lincoln Lab. Additionally, the developed methodology allows for the analysis of specific models by altitude or aircraft type. The methods presented were subsequently adjusted for a comprehensive analysis spanning across Canada's diverse airspace. The initial statistical model, derived from Canada-wide data, is currently accessible to the public via the National Research Council's GitHub repository [https://github.com/nrc-cnrc/Canadian-Airspace-Models].

Keywords

• remotely piloted aircraft systems (RPAS)
• statistical airspace model
• interacting multiple models (IMM)
• unmanned aircraft system
• Bayesian networks