Notes from our hi-res AM3 meeting at the ACCESS-NRI workshop

Atmosphere AM3
1

NRI workshop AM3 discussion.pdf (36.2 KB)
HI all, here are the notes from our joint meeting at the NRI workshop last week (can only upload pdf to the Hive, not .docx apparently). Please let me know if there’s anything incorrect or missing.

2

Discussion around resolution:

Decision between N512 / N640 to be made depending on which one is easier to achieve (neither is currently working), and on resources.

N512 is different from the Met ODice (they run N640) – could be contribution to HiResMIP2

Mat: but there is also the CABLE difference, so N640 would give a like-for like comparison of CABLE vs JULES (IF it is truly like-for-like…)

Discussion over whether to start at N216 (NRI pointed out there is no technical experience yet, even at this resolution) or the higher res. Decision to focus on higher-res (N512 / N640), and come back to N216 once hi-res is solved.

N512 missing ancils:

N512 isn’t running because it is missing the historical volcanic aerosol ancil. Julie and

Martin both say it can be run without – suggestion to turn it off and see if it runs.

N640 crash:

NRI to investigate, haven’t seen crash like this before (this late into model run). Crash happens at TOA – unlikely to be sfc variables. Christian – ozone?

Bethan – the grid point that is developing very high vertical velocities just before the crash is at high altitude and Arctic latitude, near the north pole in April. Also maybe points to ozone?

Li started from different start time, still crashes about 60-70 days in. Points to something building up on 60-70 day period, and to something in the stratosphere.

Mat - Shaun Cooper found issue in ACCESS-A ozone ancil generation when building ACCESS-A.

General science points raised:

Julie / Martin – discussion over whether the same UM version in AM3 now and planned release, also vs Met Office

Christian – tuning. Andy – no local expertise in tuning atmosphere (land yes, ocean sea ice yes). Martin – lots of cloud parameters, nobody understands what they actually are.

TODO outcomes:

  • Focus on solving N512 – probably the easier crash to fix (aerosol files).
  • Work on N216 to follow once hi-res is running.
  • Pao to work with Martin to take on the technical model side.
  • Mat, Bethan, Li on the science.
  • Create AM3 hi-res Hive group for the core group, link to it from Cumulus
  • Park Paul’s debug build for now.
  • Check the ozone ancillary files (who?)
  • Technical group (Martin, Li, Julie, Pao , Paul, anyone else?) to have a 30-min meeting every week until we can solve this (i.e. add another short meeting in the ModSci meeting weeks).
3

I have pasted the notes above for visibility. Hope that is ok.

3 Likes