Abstract
This poster reports the results of a published paper in IEEE Transactions on Visualization and Computer Graphics (TVCG) and talk presented at IEEE Visualization (VIS) 2015.
Eddies are important in climate studies because they transport heat, salt, and nutrients through the world's oceans and are vessels of biological productivity. The study of eddies in global ocean-climate models requires large-scale, high-resolution simulations. This poses a problem for timely eddy analysis, as ocean simulations generate massive amounts of data, causing a bottleneck. Future analyses at very-high spatial and temporal resolutions will be difficult with file-based workflows due to resource constraints, such as storage size and supercomputing (I/O) time. In response, we have developed an in situ eddy census workflow in MPAS-Ocean, which scales well to ten-thousand processing elements. This study demonstrates that in situ analysis has many advantages that pertain to climate model analysis tasks, as we move to higher resolutions. We report the strong- and weak-scaling performance of this workflow, along with the early science results of eddy censuses in MPAS-Ocean.