Tracking features in embedded surfaces: Understanding extinction in turbulent combustion

Understanding the temporal evolution of features of interest requires the ability to: (i) extract features from each snapshot; (ii) correlate them over time; and (iii) understand the resulting tracking graph. This paper provides new solutions for the last two challenges in the context of large-scale turbulent combustion simulations. In particular, we present a simple and general algorithm to correlate hierarchical features, embedded in time-dependent surfaces. This, for the first time, provides a parameter independent approach to track embedded features. Furthermore, we provide a new technique to adaptively change feature parameters over time to both: alleviate artifacts due to insufficient temporal resolution as well as to simplify the resulting tracking graphs to promote new scientific insights. Our solutions are integrated into a general and flexible analysis environment that allows users to interactively explore the spatio-temporal behavior of large-scale simulations. We demonstrate the results using the analysis of extinction holes in turbulent combustion as primary case study and a number of other applications to illustrate the generality of the approach.