search

ESMCI / cime

Common Infrastructure for Modeling the Earth
http://esmci.github.io/cime
Other
159 stars 205 forks source link

CPL_USER_MODS match for b1850 is too general #2282

Closed billsacks closed 6 years ago

billsacks commented 6 years ago

c8c758224a introduced this in config_component_cesm.xml:

  <entry id="CPL_USER_MODS">
    <type>char</type>
    <valid_values></valid_values>
    <default_value></default_value>
    <values match="last">
      <value compset="1850_">$SRCROOT/cime_config/usermods_dirs/b1850</value>
    </values>
    <group>run_component_cpl</group>
    <file>env_case.xml</file>
    <desc>User mods to apply to specific compset matches. </desc>
  </entry>

But matching on compset="1850_" is too general: This matches non-B compsets (e.g., I compsets). For a standalone ctsm checkout, this raises an error because $SRCROOT/cime_config doesn't exist. For other configurations, I'm not sure this is working as intended - e.g., is this supposed to be included in an I1850 compset?

@jedwards4b I'm not sure exactly what's intended here.

jedwards4b commented 6 years ago

This is just adding the orbital info for 1850, it should be a default and not set via user mods as it is here.

billsacks commented 6 years ago

Do you know what the correct behavior here is? It seems like changing this to year 1850 for all year-1850 compsets would change answers for non-B compsets... intuitively, that seems like the right thing to do, but it makes me wonder why it hasn't been done in the past.

ekluzek commented 6 years ago

Dave Lawrence just approved this change from the CLM angle for I compsets.

"I think this is fine. Should be negligible impact. "

rljacob commented 6 years ago

Are the orbital params that get used in the output somewhere?

jedwards4b commented 6 years ago

Yes, they are output in the cpl.log in lines beginning with (shr_orb_params)

rljacob commented 6 years ago

Without the usermod or the change in 2283, what is having "1850" in the compset do for you?

jedwards4b commented 6 years ago

Nothing.

jedwards4b commented 6 years ago

Sorry that isn't true. In the driver it determines what boundary files are selected, components may also set parameters based on the 1850 date in the compset.

billsacks commented 6 years ago

For example, CLM selects different datasets based on the year, and I imagine this affects CAM in many ways (CO2, aerosols, etc.) though don't know the details off-hand.