Automated flight track taxonomy for measuring benefits from performance based navigation

A flight track taxonomy is presented which decomposes a set of radar tracks according to their lateral, vertical, and conformance segments. These identifications are based upon a novel set of filtering, segment identification and track decomposition algorithms. These algorithms have been optimized such that they can batch process large data sets efficiently and robustly. Radar filtering algorithms rely upon a series of mixed nonparametric least squares filters, which are shown to isolate and minimize several sources of radar error. Next, a generalized change point analysis is described and used to identify lateral and vertical maneuvers within each radar track. Large collections of radar tracks are approximated by a reduced order module using Proper Orthogonal Decomposition (POD) and then a k-means clustering technique is applied to group these simplified tracks into common flows. In conjunction with a series of coordinate projections, this flow reduction is used measure conformance to existing procedures. In particular, deviations from nominal procedure paths due to radar vectors, direct-to clearance, and turn onto final may be identified and catalogued. This taxonomy provides numerous metrics which are valuable for measuring the benefits from performance based navigation. Two examples are demonstrated where the taxonomy is applied to the analysis of terminal descent profiles and the lateral conformance of an area navigation (RNAV) departure.