Exascale Earth System Models to Anticipate Changes in a Warmer World

Exascale Earth System Models to Anticipate Changes in a Warmer World

Monday, May 22, 2023 3:00 PM to Wednesday, May 24, 2023 5:00 PM · 2 days 2 hr. (Europe/Berlin)
Foyer D-G - 2nd Floor
Project Poster
Climate and Weather ModelingDigital TwinsExascale SystemsManaging Extreme-Scale ParallelismParallel Programming Languages

Information

With modern HPC machines becoming increasingly diverse in architecture and vendors (mainly GPUs from Nvidia, AMD and Intel, but also niche systems like NEC vector or Fujitsu A64FX), the numerical and climate modeling community is facing huge challenges to adapt to these systems in the exascale era. However, at the same time these systems also offer exciting new possibilities by e.g. allowing for the first time to do global climate simulations on km-scales, which enables the direct simulation of important processes like storms and ocean eddies. In this poster, we present WarmWorld, a German national project that aims to harness advances in information technology to compute and evaluate climate warming trajectories and make their inherent information content transparent to scientific communities. A central component of this climate information system will be an ICON-based earth system model capable of resolving global oceanic and atmospheric coupled circulation systems on kilometer-scales with a throughput of one simulated year per day on future exascale computing systems. To achieve this climate scientists and software engineers will work together in four modules: “Better”, which is responsible for defining, improving and testing the model configurations. “Faster”, which is responsible for transforming the ICON code base into an open, scalable, modularized and flexible code named ICON-consolidated. “Easier”, which will develop novel methods to make climate information visible, accessible, and interoperable. “Smarter”, which aims to involve the applied math and informatics communities to improve the workflow and the model performance.
Contributors:
Format
On-site
Beginner Level
30%
Intermediate Level
50%
Advanced Level
20%