ICON Meeting 2026/3¶
1 October 2026, ETH Zurich (D 26.2) and via Zoom
Participants (on-site)¶
Michael Jähn (MJ), Matthieu Leclair (MaL), Alitzel Macías (AM), Marius Rixen (MR), Christian Zemman (CZ), Prince Asilevi (PA), Michael Steiner (McS)
Participants (Zoom)¶
Mikael Stellio (MS), Alina Yapparova (AY), Tina Schnadt (TS), Christian Steger (CS), Matthias Kraushaar (MK), Clarissa Alicia Kroll (CK), Marco Arpagaus (MA), Andrea Stenke (AS), Christoph Riess (CR), Doris Folini (DF), Jacopo Canton (JC)
Minutes by Alitzel Macías
Reports¶
C2SM (Michael Jähn, Matthieu Leclair, Mikael Stellio, Alitzel Macías)¶
MJ mentions that to receive the invitation for this meeting, one needs to subscribe to the C2SM ICON mailing list on the C2SM landing page.
MJ welcomes everyone and shares the latest news about ICON from C2SM .
CZ asks about the reason behind the name --gpus-per-node in the recent node sharing changes on Säntis; MK replies that it was the most convenient choice. He also confirms that --gres-flags=allow-task-sharing must be passed to srun (not in the #SBATCH block). CSCS is still working on the accounting, so that users only get billed for the GPUs they actually use; this is expected to be ready by the end of the year.
MaL mentions that when selecting NUMA nodes, one has to explicitly ask the system which socket was assigned, to make sure that the binding is correct.
CK asks where to report issues with ICON, as this is currently not very clear. MaL answers that it will get better once the repositories are unified on icon/icon. TS suggests flagging issues internally first and letting C2SM report them, since otherwise no one is clearly responsible. MaL says that this needs to be discussed further, especially given the upcoming switch to nightly-only CI testing. CK's specific issue only occurred with 24 nodes, which is why the bug was not caught by the CI.
Regarding the issue with the HWSD (Harmonized World Soil Database) soil data in EXTPAR, CZ says that he has a dataset that MS can compare with the current one to evaluate its correctness. MA is also interested in this dataset from CZ.
Christian Zemman¶
Working on European high-resolution simulations (ensemble, 3 km, 30 years of current climate and 30 years of future climate at 3 degrees of warming), which is a multi-year project.
Prince Asilevi¶
Working at Empa, setting up ICON over West Africa. He will be looking at chemistry simulations; the project duration is 5 years.
Michael Steiner¶
Also working on chemistry simulations over Africa, focusing on aerosol emissions, methane and satellite data. He is using ICON-ART on Eiger (CPU) together with the HWSD soil dataset. He will have some future requests for C2SM: porting code to GPUs and Zonda inputs for urban modeling.
Christoph Riess¶
Also working on ICON-ART for methane inversions, on Säntis and at European scale.
Andrea Stenke¶
Working on ICON-ART for simulating atmospheric trace elements (selenium, arsenic, etc.) on Euler.
Doris Folini¶
No update.
Clarissa Alicia Kroll¶
Working on the implementation of an extension to the damping scheme in the PBL, using 5 km simulations.
Marco Arpagaus¶
Working on a new version of the radiation-topography corrections, 3D radar, and homogenizing setups. They are also working on ATP, where there are some issues with the model being too dry.
Christian Steger¶
Also working on the radiation-topography scheme. It is now considerably faster than before, and both the grid-scale and subgrid-scale versions should be available soon.
Jacopo Canton¶
Working within EXCLAIM on ICON4Py (see also the news post on the ICON4Py dynamical core). He would like input from the ICON community on which features would be useful in ICON4Py. CZ says that an overview would be a good start.