Ensemble-Vis: A Framework for the Statistical Visualization of Ensemble Data Kristin Potter * , Andrew Wilson † , Peer-Timo Bremer ‡ , Dean Williams ‡ , Charles Doutriaux ‡ , Valerio Pascucci * , and Chris R. Johnson * * Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, Utah 84112 † Sandia National Laboratories, Albuquerque, New Mexico, 87185 ‡ Lawrence Livermore National Laboratory, Livermore, California, 94550 Email: {kpotter,pascucci,crj}@sci.utah.edu, atwilso@sandia.gov, {bremer5,williams13,doutriaux1}@llnl.gov Abstract—Scientists increasingly use ensemble data sets to explore relationships present in dynamic systems. Ensemble data sets combine spatio-temporal simulation results generated using multiple numerical models, sampled input conditions and perturbed parameters. While ensemble data sets are a powerful tool for mitigating uncertainty, they pose significant visualization and analysis challenges due to their complexity. In this article, we present Ensemble-Vis, a framework consisting of a collection of overview and statistical displays linked through a high level of interactivity. Ensemble-Vis allows scientists to gain key scientific insight into the distribution of simulation results as well as the uncertainty associated with the scientific data. In contrast to methods that present large amounts of diverse information in a single display, we argue that combining multiple linked displays yields a clearer presentation of the data and facilitates a greater level of visual data analysis. We demonstrate our framework using driving problems from climate modeling and meteorology and discuss generalizations to other fields. Index Terms—Ensemble data, uncertainty, statistical graphics, coordinated and linked views. I. I NTRODUCTION Ensemble data sets are an increasingly common tool to help scientists simulate complex systems, mitigate uncertainty, and investigate sensitivity to parameters and initial conditions. These data sets are large, multidimensional, multivariate and multivalued over both space and time. Due to their complexity and size, ensembles provide challenges in data management, analysis, and visualization. In this article we present Ensemble-Vis, a general framework to support the visual analysis of ensemble data with a focus on the discovery and evaluation of simulation outcomes, a screenshot of which can be seen in Figure 1. Our approach combines a variety of statistical visualization techniques to allow scientists to quickly identify areas of interest, ask quan- titative questions about the ensemble behavior, and explore the uncertainty associated with the data. By linking scientific and information visualization techniques, Ensemble-Vis provides a cohesive view of the data that permits analysis at multiple scales from high-level abstraction to the direct display of data values. Ensemble-Vis is developed as a component-based framework allowing it to be easily adapted to new applications and domains. Fig. 1. The Ensemble-Vis framework provides a platform for data visualiza- tion and analysis through a combination of statistical visualization techniques and a high level of user interaction. A. Motivation The goal of an ensemble data set is to predict and quantify the range of outcomes that follow from a collection of simula- tion runs. These outcomes have both quantitative aspects, such as the probability of freezing rain in a given area over a given time, and qualitative aspects, such as the shape of a severe weather system. While ensemble data sets have enormous power to express and measure such conditions, the appropriate methods for visualization and analysis are highly dependent on the application area. We focus on two driving applications: short-term weather forecasting and climate modeling. Meteorologists use short-term forecast data to predict local weather outlooks rather than relying on singular, deterministic models [1]. These data sets give insight into the range of possible predictions and allow meteorologists to provide weather forecasts along with the probability of particular outcomes. We use data from NOAA’s Short-Range Ensemble Forecast (SREF), obtained from the National Centers for Environmental Protection’s Environmental Modeling