Presentation
Visualization in Earth System Research - Current Status and Challenges imposed by HPC Trends
SessionVisualization and HPC
Speaker
Event Type
Focus Session
Big Data Analytics
HPC workflows
Visualization & Virtual Reality
TimeMonday, June 25th5pm - 5:20pm
LocationPanorama 3
DescriptionBy utilizing world class HPC systems, numerical models of the entire Earth system are used to simulate the climate of the past, of today and potential future developments. Currently, results of such computations are stored at regular time intervals in the form of multi-variate three-dimensional fields, which then need to be analyzed. Although time periods of up to tens of thousands of years are simulated, the storage intervals have to be short enough to enable analysis of short-term variations such as, e.g., the daily cycle. In this workflow, visualization plays a vital role: first, it is used to comprehend the results. Second, visualizations are indispensable for communicating the findings within the scientific community and to the public.
However, the computational performance of HPC systems grows faster than their I/O speed, and this requires a paradigm change of the scientific workflow. Today, we are scratching the limits of what is technically possible without in-situ or other technologies that enable visualization of very large scientific data. Furthermore, ensemble simulation techniques are becoming more widely used, and the robustness/uncertainty dimension has to be taken into account for visualization and communication of model results.
By showing many examples from practice at DKRZ, I will give an overview on the current the state-of-the art in the visualization of climate model data. Finally, I will discuss upcoming challenges and future strategies.
However, the computational performance of HPC systems grows faster than their I/O speed, and this requires a paradigm change of the scientific workflow. Today, we are scratching the limits of what is technically possible without in-situ or other technologies that enable visualization of very large scientific data. Furthermore, ensemble simulation techniques are becoming more widely used, and the robustness/uncertainty dimension has to be taken into account for visualization and communication of model results.
By showing many examples from practice at DKRZ, I will give an overview on the current the state-of-the art in the visualization of climate model data. Finally, I will discuss upcoming challenges and future strategies.
Speaker
