The Ruby project bundles MPI-wide activities invoking ICON-ESM in climate mode. Specific to this mode is the usage of 'parameterized physics' for small-scale processes instead of resolving them explicitly at their characteristic scale. By the resulting speedup, the Ruby set of ICON models is tailored for global studies aiming at at-least century-scale simulations.

Ruby experiments include paleo-climate studies with exotic continental configurations, large ensembles of historical and climate-change simulations, and investigations of small-scale ocean-atmosphere interactions and their impact on decadal to multidecadal variability.

ICON Configuration ICON-Ruby (ICON-ESM)


The Sapphire project develops an ICON version where, in contrast to state-of-the-art General Circulation Models (GCM), the major modes of energy transport are resolved explicitly. This requires a grid spacing of a few kilometers to capture deep convective eddies in the atmosphere and mesoscale eddies in the ocean, with an ambition for hectometer-scale modeling to resolve shallow convection and sub-mesoscale ocean eddies. Whereas such storm-resolving models (SRMs) are now well established for regional climate simulations and weather forecasts, their global application is still in its infancy given the considerable computational costs associated with such simulations. The same statement applies to sub-mesoscale dynamics in the ocean that were studied either with idealized or limited-area experiments but not in global configurations. As an example, a 5-km global coupled ICON simulation needs one day (in real time) to simulate one month using one sixth of the supercomputer at DKRZ, whereas a similar simulation with a grid spacing of 100 km would only need a few seconds.

The Sapphire project is organized around experiments (see Tab. 1) that both serve to advance our scientific understanding and the technical abilities of ICON. As global storm-resolving simulations remain rare, they generally investigate uncharted territory. One big scientific unknown in this respect concerns the importance of the newly resolved small-scale features of the atmosphere and of the ocean for the large-scale circulation of the atmosphere and of the ocean, as well as for setting mean properties of the climate system. On the technical side, the Sapphire experiments are used to successively update the physical parameterizations of ICON, making them more adapted for high-resolution modeling, and to investigate new coding paradigms, such as GPU, to achieve a higher performance.

Table: List of conducted Sapphire experiments. Only closed experiments are listed.

Experiment name



Regional simulation of the monsoon during Holocene at 5 km


Simulate 40 days (1.8-10.9.2016) globally at 2.5 km, see Hohenegger et al. (2020)


Simulate the coupled climate system for 5 months at 5 km


Simulate the winter climate (EUREC4A period) globally, at 2.5 km (uncoupled) and 5 km (coupled)


Port the ICON model to GPU to be able to simulate a QBO event with 200 vertical layers and 2.5 km horizontal grid spacing


Simulate sub-mesoscale processes in global ocean using a refined grid spacing over the North Atlantic from 11 km down to 600 m

ICON Configuration ICON-Sapphire (ICON-LEM)