RT Journal Article ID 48b130c32c908cf5 A1 Liu, Zhanping A1 Moorhead II, Robert J. T1 HIGH-PERFORMANCE FLOW VISUALIZATION FOR EFFECTIVE DATA ANALYSIS JF Journal of Flow Visualization and Image Processing JO JFV YR 2016 FD 2017-02-21 VO 23 IS 1-2 SP 41 OP 57 K1 scientific visualization K1 flow visualization K1 parallel visualization K1 line integral convolution K1 streamlines K1 pathlines K1 streamline placement AB As data grows at exponential rates toward exa-scale (1018) computing, it is necessary to exploit scientific visualization in the form of a visual, informative, and interactive methodology to help resolve daunting problems arising from a wide variety of disciplines that involve big data analysis. While novel visualization algorithms are investigated for effective as well as efficient exploration of surface and volume flows, there are signs of revisiting sparse geometry-based methods with further improvement, back from dense texture-based approaches, and resorting to parallel visualization. This paper presents our innovative research along this path in high-performance visualization of flow data for exploration, recognition, representation, and analysis of overall patterns and salient features, with techniques from texture-based to geometry-based, flows from 2D to 3D, complexity from steady to unsteady, and computing from serial to parallel. Specifically, a high-level description is primarily focused on four algorithms, i.e., accelerated unsteady flow line integral convolution, its extension to time-varying volume flows, advanced evenly spaced streamline placement, and interactive view-driven evenly spaced streamline placement. Also introduced are four systems, i.e., ActiveLIC, ActiveIBFV, ActiveFLOVE, and DOXIV, which we developed for high-performance flow visualization coupled with applications to demonstrate the practical applicability. PB Begell House LK https://www.dl.begellhouse.com/journals/52b74bd3689ab10b,62381c570f9a8280,48b130c32c908cf5.html