Clustering radar tracks to evaluate efficiency indicators

Abstract

Clustering radar tracks to evaluate effectiveness indicators It is expected that air traffic increases considerably in the next years whereas the possibility to extend the available infrastructure is limited. Therefore, in future the infrastructure has to be used more efficient than today. The airspace surrounding an airport is one of the resources that can not be extended like the number of runways (which is hard enough). On the contrary, in highly populated areas like London, airspace is a valuable resource that is needed to separate approaching aircraft. In this paper, we analyze approaching aircraft within the last 100NM of an airport to compare different approach strategies. The efficiency of an approaching aircraft separation strategy is measured by calculating efficiency indicators which consist basically of three values for the last 100NM: the flight time median, flight time variance, and an idealized flight time. The last indicator takes the complexity of the airspace into account. The ideal flight time represents the time an aircraft typically needs to cross the last 100NM if it is not disturbed by other aircraft. To obtain useful values of the three presented indicators, approaching aircraft are clustered into groups. In other words, the (unsorted and unstructured) radar tracks must be sorted into groups in which they are comparable: it makes no sense to analyze the flight time for aircraft that approach the airport from different directions due to the differing approach routes. The clustering can be done by two general algorithms which are the main focus of this paper: clustering by origin and runway configuration respectively and clustering by radar track similarity. Both clustering approaches can be further subdivided by using different clustering methods. In this paper, we present the clustering methods as well as compare their results. The analysis is done solely on radar tracks, no information on airport structure, airspace structure or separate runway configuration is used. Also the clustering technique is designed to hold minimal human interaction which allowes automated analysis of many airports.