Immersive Visualization of The Multifaceted Uncertainties of Hurricane Prediction Ensembles

Visualizations play a crucial role in interpreting hurricane forecasts and facilitating emergency decision-making. However, existing technologies have limitations in effectively conveying the multifaceted uncertainties included in these forecasts. These uncertainties include the spatial spread of forecast tracks, forecasting time, and forward speed. The visualization of hurricane prediction tracks in itself presents a high level of complexity. Incorporating additional visual encoding to represent uncertainties can exacerbate this complexity, making it even more challenging to interpret. To address these challenges, we introduce an immersive uncertainty visualization using stereoscopic rendering. Individual hurricane tracks are rendered as color-coded textured tubes, where color represents forecasting time, and the texture dynamically changes to represent the uncertainty of forward speed. To prevent visual clutter from overlapping tracks, we map time-specific predicted locations along these tracks to varied heights, thereby enhancing depth perception and facilitating an improved interpretation of the visual encodings. We conducted an experimental study to evaluate the effectiveness of our approach, examining participants' estimations of the distributions of these uncertainties. The results reveal that participants were generally successful in interpreting these time-specific uncertainties, and exhibited greater accuracy in estimating distributions of high and low forward speeds compared to medium speed.