Logo Logo
Help
Contact
Switch Language to German
Kolb, Daniel ORCID: 0000-0001-9645-3319; Kurtz, Wolfgang; Weismüller, Jens; Ramm, Alexander von; Ludwig, Ralf ORCID: 0000-0002-4225-4098; Kranzlmüller, Dieter (5. September 2018): Visualization of climate simulation data in virtual reality using commercial game engines. Enviroinfo 2018, September 5th - 7th 2018, Garching near Munich, Germany.
[img]
Preview
Creative Commons Attribution - Accepted Version 414kB

Abstract

Due to the size of its customer base the video game industry has long been the best-funded proponent of innovative real-time computer graphics. Many advancements in the field of computer graphics, software and hardware, have become cost-effective due to their use in video games, which in turn funded even further research and breakthroughs. Recent changes in the monetization of commercial game engines made their use in less revenue driven institutions affordable and, hence, possible. This allows us, given suitable hardware, to build and run computationally expensive fully interactive real-time visualizations at a fraction of the cost and time. We can thus investigate and explore the data in our virtual reality application far sooner. Additionally, we are able to spend more time to iteratively refine the user interaction as well as the preprocessing of the raw scientific data. We supply our visualization with the output data of ClimEx’ computational run on the SuperMUC. ClimEx is a research project that studies the effects of climate change on meteorological and hydrological extreme events. It features a multitude of climate-relevant variables and observes the time span between 1950 and 2100. For our use case we chose to compare three different precipitation events. Each event consists of simulated 60 hours of rainfall data anteceding a potential 100-year flood, which is a flood event that has an annual exceedance rate of 1%. The first event draws from historical data and represents the rain leading up to the 1999 Pentecost flood. We compare these data with two computer generated prospective events, which take place in 2060 and 2081, respectively. Since we wish to gain knowledge on strong local extrema as well as the comprehensive overall trend of the attributes, we chose to display the data in virtual reality. The virtually unlimited number of perspectives and points of view simplify investigating and understanding the three-dimensional data. We are also able to place the observer at the center of the data and empower them to interact with and steer the visualization in intuitive ways. By utilizing a tool like virtual reality, we are able to create an immersive, interactive and engaging user experience, which further facilitates the user’s ability to focus on the visual display and extract information from the displayed data. This allows users, especially non-expert users, to grasp the data we present in our visualization with less effort. In our paper we present the necessary steps to create an immersive virtual reality 3D visualization from raw scientific data based on our use case. This entails several aspects of pre-processing, a simple, suitable user interface as well as our solutions to the challenges we encountered.