Similarity search of spatiotemporal scenario data for strategic air traffic management

Designing strategic air traffic management (ATM) solutions at a large spatiotemporal scale in real time is challenging, considering the range of uncertainties at the strategic time frame. Big data techniques have drawn increasing attentions to develop optimal ATM solutions to address these challenges. ATM data, such as convective weather spread and congestion propagation, represent a new data type called spatiotemporal scenario data, which has not been systematically studied. This new data type differs from the traditional spatiotemporal pointwise data in its unique spatiotemporally correlated spread patterns. As a step toward closing the loop of big data and real-time decision making for ATM, this paper introduces an effective similarity search algorithm for this new data type. This similarity search algorithm uses a multiresolution distance measure, which captures the difference between spatiotemporal scenarios. Unique properties of this distance measure are exploited to significantly reduce the computational cost associated with accessing and processing scenarios in a large database. Using real weather forecast datasets as the case study, this study investigates feasibility of the proposed similarity search algorithm. Systematic parameter impact analysis is conducted through simulation studies, which provide guidelines for parameter selection. Comparative simulation studies validate the effectiveness and efficiency of the proposed similarity search algorithm for spatiotemporal scenario data.