Real-time Interactive Exploration of Large Atmospheric Datasets in Virtual Reality

Abstract

While technological advances in high performance computing allow for an ever-increasing accuracy in climate and weather simulations, they also lead to grand challenges regarding the data visualization and analytics process. We present a visualization framework, which allows for interactive exploration and real-time visualization of such large scale datasets in virtual reality. It combines parallel and distributed feature extraction using high-performance computing resources with octree-based level-of-detail rendering methods to assure high frame rates during the complete analysis process. When parameters such as an iso-value or the current time-step are modified, the visualization is updated progressively in a view-dependent manner. In addition, the data is shown in relation to the geographical, planetary and celestial context: the data is shown as part of our solar system. Planets are rendered with a sophisticated level-of-detail system based on the HEALPix tessellation of spheres. HEALPix tiles have equal areas and do not suffer from singularities at poles which are a common issue with other tessellations. Geographical datasets (e.g. Satellite images, digital elevation data or vector maps) are loaded from Web-Map-Services (WMS). Example datasets for Earth include, but are not limited to, Sentinel images, Open Street Map, TanDEM-X or SRTM30. NASA's SPICE library is used to compute the position of sun, planets, moons and stars.