Replacing the YATM atmosphere in ACCESS-OM2 with intermediate complexity model SPEEDY

COSIMA
, , , ,
1

Hi COSIMA team,

I’m returning to work that I started in 2024, the initial part was related to porting and running ACCESS-OM2 on Leonardo supercomputer in Italy under SLURM scheduler.

As the next step I’m starting to replace YATM driver with an active atmosphere - the SPEEDY intermediate complexity model to run long experiments for ocean sensitivity. A similar approach has been done for NEMO ( SPEEDY-NEMO: performance and applications of a fully-coupled intermediate-complexity climate model | Climate Dynamics | Springer Nature Link )

I would be grateful for any feedback on this plan and any comments related to the technical aspects of coupling SPEEDY with ACCESS-OM2.

While looking for reference implementations within the ACCESS, I came across the fully coupled configuration ACCESS-CM2, which uses the Unified Model (UM):

http://climate-cms.wikis.unsw.edu.au/ACCESS-CM_2.0

I realise that parts of ACCESS-CM2 are not openly available. I wanted to ask, in a general sense, whether the source code or example coupling/configuration files for ACCESS-CM2 are accessible to users, or if this setup is considered closed?

Many thanks in advance!

Natalia

2

Thanks for the question, it’s a tricky one.

That’s right - the CM2 configurations aren’t available publicly at this point - although the ocean and sea-ice model components are somewhat shared with OM2.

You might find the ACCESS-ESM1.6 configurations somewhat useful, although the atmosphere model (UM) is not open source.

ESM1.6, like OM2, also uses MOM5 and CICE5 although the coupling with the atmosphere is quite different to the coupling with the data atmosphere in OM2. From the atmosphere to sea-ice, OM2 couples states like air temperature, humidity and wind and the sea-ice model calculates sea ice surface temperature and heat fluxes. Whilst in ESM1.6, the atmosphere model calculates sea ice surface temperature and heat fluxes, and then heat fluxes and wind stress is coupled to the sea-ice . I think the fields coupled in the OM2 approach is probably more common, but the coupling is simplified because there is no feedback from the ocean/sea ice to the atmosphere.

It looks like OM2 and ESM1.6 use the OASIS3-MCT coupler, and it looks like SPEEDY-NEMO, also has been used with OASIS3-MCT. I guess one thing to look at would be the fields coupled with the atmosphere and compare those in SPEEDY with those use in OM2.

3

Hi Anton,

Thank you! I’ve now realised that the key thing is to work with coupled fields and rather replicate ESM strategy but with SPEEDY.

I’ll keep this post updated.

1 Like
4

I suspect a hybrid approach is needed, but it depends somewhat on what fields SPEEDY is calculating / expecting to be coupled. ESM 1.x all use sea ice surface (skin) temperature calculated in the atmosphere, but OM2 prescribes skin temperature from the data forcing. So if SPEEDY is expected sea ice surface temperature to be calculated in the sea ice model (which is probably simpler / more accurate) then it’s a hybrid of the two approaches.

1 Like