2016-06-07 Coupled Tasks Meeting notes

Owned by
Peter Caldwell
Last updated: Jun 08, 2016 by
Milena Veneziani
3 min read

Date

Time

  • PST:
  • EST: 3:40-5:40pm

Call Info

  • web session: 
  • call number:

Attendees

  • lots of people

Goals

What Actually Happened

  • Chris Golaz read through the tasks listed under goals.
    • Robert Jacob expressed surprise that he was in charge of water+energy budgets (reflecting communication breakdown)
    • Peter Thornton noted that land tuning task was related to leaf albedo rather than leaf area index
  • Peter Thornton: does each component have its own conservation checking?
    • Phil Rasch (pnl.gov): atmos has checks
      • Peter Caldwell: but they are broken and we know from RESTOM-RESSURF and other checks that energy and water in atmos don't conserve.
    • Doug Jacobsen (Unlicensed): ocean has conservations checks. Problem seems to be in coupling between components - he has a python script available to look at coupler history files. He figured out part of the reason why ocean lost 3m was river runoff was screwed up. Still have ~1m of lost ocean over 40 yrs though. Notes that G cases conserve... only B cases don't.
    • Gautam Bisht notes that CLM4.5 doesn't crash when energy conservation errors are found (error call is commented out). It should print warning messages though.
  • Phil Rasch (pnl.gov) points out that he's still unclear whether snowfall onto land or ocean accounts for the latent heat of fusion properly
    • Doug Jacobsen (Unlicensed) assures us that ocean is carrying latent heat of fusion for snow and ice (i.e. snow melts and takes up heat when hits ocean)
  • On a related note, Doug Jacobsen (Unlicensed) points out that the ocean assumes river runoff occurs at whatever temperature the ocean layer it flows into has.
    • Todd Ringler (Unlicensed) points out that there are lots of places like this where one component makes assumptions about transfer across components because the info it needs isn't available through the coupler (or at all).
      • This makes a small effect globally (O(0.1 W/m2)), but can be big regionally (O(10 W/m2)).
      • Chris Golaz points out that passing rain T to ocean would require tracers tracking the temperature of ice, liquid, etc in each cell. This would be a big undertaking.
  • David C. Bader pointed out the need to do the CMIP Deck experiments (see standard_experiments.pdf)
    • noted we will do all at low res (i.e. 1 degree)
    • noted we need compsets for each of these
      • Philip Cameron-Smith (Unlicensed) asks whether we can use CMIP5 input data for our compsets, or whether we need to update to CMIP6
        • Chris Golaz says it would be good to stay as modern as possible
        • David C. Bader and others say the biggest priority is to get something working rather than shooting for perfection
    • Peter Thornton asked when deck experiments need to be done. Dave:June 30th of next year
      • Mark Taylor - how many hours do we need
        • Chris Golaz (after calculations) = ~50M core hours on Edison
      • Mark Taylor notes that if we bought our own dedicated machine, it would probably only supply ~30M core hours/yr.
    • Peter Caldwell - When do we need to start these runs in order to finish on time?
      • Mark Taylor - we did 50M hrs on Edison in last 5 mo, so about that much time?
      • Ruby Leung - since CMIP6 is delayed, perhaps we can move our own timeline back?
        • David C. Bader - our timeline is set by ACME deliverables rather than CMIP6, so no. By review we absolutely need to have a water cycle experiment running (but not finished) and completed AMIP, PI, and historical runs completed at low res.
  • David C. Bader asks how many people know how to make mapping files
    • Doug Jacobsen (Unlicensed) - there are instructions on confluence which anybody could use to create mapping files in ~1 hr
      • David C. Bader - but there's always something that goes wrong and requires by-hand effort to fix. How many people know how to fix this?
    • Stephen Price - making the mapping files more robust has been on our to-do list. Should we elevate?
      • yes.
  • David C. Bader: what is the status on coupled diagnostics
    • Kate Evans (Unlicensed) - basics are done for the atmosphere. ORNL has been running diagnostics for Chris Golaz but would like to have him run the code. 
    • Milena Veneziani - basics are done for ocean/sea ice diagnostics. Working now on integrating python scripts into the coupled-diags shell script that Salil Mahajan wrote for Chris Golaz and others to use. Goal now is to have something working asap for edison/cori, where simulations are currently being run.

Action items

  •