Swarm Data Mining for the Fine Structure of Thermals

Abstract

Accurate thermal models can help to optimize the design of sailplanes. Theoretical models should be based on empirical evidence. However, very few measurements on the structure of thermals are published. This paper uses data mining techniques on data collected by swarms. In this case the swarm consists of the world’s best pilots in the world’s best gliders competing in a world championship at Uvalde, Texas in 2012. It is pointed out how the data collected by this swarm in the form of ICG files (i.e. GPS recordings) may be processed in order to yield the vertical speed of the air in thermals. This resulted in about 100 hours of data on thermals. From this data a model of the fine structure of thermals could be derived consisting of three components: a Gaussian representing the buoyancy, a vortex modeling entrainment and a border vortex caused by the difference in speed between the air inside the thermal and the surrounding air.