Visualizing events in time-varying scientific data

In time-varying scientific datasets from simulations or experimental observations, scientists always need to understand when and where interesting events occur. An event is a complex spatial and temporal pattern that happens over a course of timesteps and includes the involved features and interactions. Event detection allows scientists to query a time-varying dataset from a much smaller set of possible choices. However, with many events detected from a dataset, each spanning different time intervals, querying and visualizing these events pose a challenge. In this work, we propose a framework for the visualization of events in time-varying scientific datasets. Our method extracts features from a data, tracks features over time, and saves the evolution process of features in an event database where a set of database operations are provided to model an event by defining the stages or individual steps that make up an event. Using the feature metadata and the event database, three types of event visualizations can be created to give a unique insight into the dynamics of data from temporal, spatial, and physical perspectives and to summarize multiple events or even the whole dataset. Three case studies are used to demonstrate the usability and effectiveness of the proposed approach. Graphic abstract