Spatio-Temporal Mapping -A Technique for Overview Visualization of Time-Series Datasets-

Numerical simulations have recently increased in scale and have often output high dimensional (three-dimensional and time-evolving) datasets. This makes it difficult for users to quickly grasp physical phenomena involved in such datasets. To overcome this difficulty, we propose a spatio-temporal information mapping technique with a map-design capability (spatio-temporal map) by using an information visualization technique. The spatio-temporal map is generated by mapping the values of a three-dimensional and time-evolving physical quantity into a two-dimensional space with spatial and temporal axes. Here, three-dimensional spatial information is condensed into one dimension by subdividing a target model with an octree. By using the map, users can quickly find regions of interest involved in high dimensional datasets. In addition, users can interactively change several aspects of the map such as its resolution and color coding method. Furthermore, users can design the map by changing the tree structure. By applying the spatio-temporal map to a full-scale three-dimensional vibration simulator for an entire nuclear power plant, we confirmed that the map is a useful technique to quickly identify appropriate regions of interest.