Interactive exploration of flow features in large-scale 3D unsteady flow data is one of the most challenging problems in visualization today. In an effort to comprehensively explore the complex feature spaces in these datasets, we have designed a scalable framework for investigating a multitude of characteristics from traced fieldlines. This new capability has allowed us to examine various neighborhood-based geometric attributes in concert with other scalar quantities, a type of analysis that

more »

... s not previously possible due to the large computational overhead and I/O requirements. We have integrated visual analytics methods into our approach by allowing users to procedurally and interactively describe and extract high-level flow features. In this work, we show the generality and expressiveness that is offered by this approach, and we show its efficacy by exploring various phenomena in a large global ocean modeling simulation.