Opened 5 months ago

Last modified 9 days ago

#367 accepted

CABLE PIcontrol suite

Reported by: Martin Dix Owned by: Martin Dix
Priority: major Component: ACCESS-CM2
Keywords: Cc:


u-bf481 is the candidate coupled model PPI control suite with CABLE. This ticket documents the configuration set up and spinup.

Change History (10)

comment:1 Changed 5 months ago by Martin Dix

Owner: set to Martin Dix
Status: newaccepted


u-ax871 started from JULES PI control u-at658 at model time 02010101. D2 iceberg scheme. CABLE code /home/599/mrd599/mosrs/jules/r11864_tvrad_fix_ih_merge. This was a manual merge of branches/dev/ianharman/vn10.6_CABLE_r11864@12302 and branches/dev/martindix/r11864_tvrad_fix@12260.

u-ba648 started from u-ax871 03010101. Configuration as for u-ax871 but with ocean SW changes and CABLE changes

  • Medlyn stomatal conductance
  • PFT dependent root depth
  • Albedo calculated with correct zenith angle rather than timestep lagged.

CABLE code /branches/test/martindix/r12417_albedo_fix, branches/dev/Share/vn10.6_CABLE@12579

u-bb194 started from u-ax871 03010101. Configuration as for u-ba648 but with CABLE changes

  • Antarctic & Greenland snow albedo increased to NIR=0.65, VIS=0.9.
  • Increased permanent snow roughness
  • Vegetation convex parameter

CABLE code is branches/dev/Share/vn10.6_CABLE@13036.

u-bd052 started from u-ax871 03010101. Configuration as for u-bb194 but using NCAR LAI, /projects/access/data/ancil/access_cm2_n96e/O1/cable_vegfunc_ncarlai.anc. CABLE code branches/dev/Share/vn10.6_CABLE@13036

u-be106 started from u-bd052 04010101. Configuration as for u-bd052 but with ocean albedo scaling set to 1.02. CABLE code branches/dev/Share/vn10.6_CABLE@13036, branches/dev/martindix/vn4.7_access_cm2@12754. Note that the latter branch is just to handle the albedo scaling namelist variable.

Restart reproducibility experiments with u-be106 showed a sensitivity to the timestep number. This was traced to CABLE incorrectly scaling precipitation rates by timestep number rather than timestep length in one part of the code.

New suite u-bf481 with this fixed started from u-be106 05010101.

CABLE code branches/dev/martindix/r13036_precip_scaling@13918 branches/dev/martindix/vn4.7_access_cm2@12754

Last edited 5 months ago by Martin Dix (previous) (diff)

comment:2 Changed 5 months ago by Martin Dix

History of u-bf481

At start of year 578 switched to the harmonised ocean code,

Ozone ancillary was being updated every 30 days and LAI, DMS and chlorophyll every 5 days. These settings are not appropriate with the Gregorian calendar and affect reproducibility of NRUN restarts. Ozone changed to monthly updating and others to daily at start of year 691. Expect a negligible effect on runtime and results.

Minor suite changes that don't affect results

Reduced UM output gives a significant speedup (~15%)

Make gregorian calendar a fixed setting rather than an option

Specify CICE revision for build

Convert CICE output to netCDF (change in driver scripts). This revealed a CICE error that meant daily history wasn't saved for leap days. Fixed in r402 of cice_gsi8.1 branch. Also apply fix for the CICE error handling.

Last edited 4 months ago by Martin Dix (previous) (diff)

comment:3 Changed 4 months ago by Martin Dix

Candidate PI Control CMIP6 STASH from Roger's u-bf022.
Stash Table

Export STASH from u-bf022 using

cd ~rb4844/roses/u-bf022/app/um
rose macro stash_copy.STASHExport

In target suite, use rose-edit to remove the existing STASH and then run the metadata StashProfilesRemoveUnused? macro. Outside rose-edit

rose macro stash_copy.STASHImport

Change domain profiles dsoil to 6 levels, dpft to 13 and dtile to 17 for CABLE.

Add back the CABLE STASH package from the original suite plus a few extra tiled variables.

Set all files to reinitialise every calendar month.

MOM diagnostics changed to save conservative temperature as well as potential temperature and to set diagnostic SST to potential temperature. Also extra daily diagostics and other OMIP diagnostics requested by Abhi.

Run with these diagnostic changes restarted at 07000101.

Last edited 4 months ago by Martin Dix (previous) (diff)

comment:4 Changed 4 months ago by Martin Dix

FAFMIP setup changed results so not using this at the moment.

Last edited 4 months ago by Martin Dix (previous) (diff)

comment:5 Changed 4 months ago by Martin Dix

As requested by Arnold, change CICE diagnostics from




Change applies from the start of year 719.

comment:6 Changed 4 months ago by Martin Dix

Extra ocean diagnostics from Abhi. Extra UM diagnostics for energy fluxes over sea ice and land tiles. Also removed 8,202 LAND SNOW MELT HEAT FLUX because it's not saved correctly by CABLE. It can be calculated from 8,231 LAND SNOW MELT RATE.

Changes apply from start of year 724.

Added ice category fields, SH and LH fluxes over open sea. Also tropopause height and density for ozone re-distribution.

Changes apply from start of year 727.

Last edited 3 months ago by Martin Dix (previous) (diff)

comment:7 Changed 3 months ago by Martin Dix

Extra ocean diagnostics from Abhi saved from start of year 770.

comment:8 Changed 3 months ago by Martin Dix

CMIP6 requires extra diagnostics for the PI control period equivalent to 1985-2014. This is model years 835-864. These were copied from Roger's u-bd266 and added with a package name PIextra to make it simple to turn them off again later.

Model warm-restarted at year 835 with these changes.

comment:9 Changed 3 months ago by Martin Dix

Restarted again at 865 with the PIextra package turned off.

comment:10 Changed 4 weeks ago by Martin Dix

Found a problem with diagnostic 10m wind speed that affected sea-salt aerosol. New run with this and other sea-salt aerosol fixes u-bi889 started from year 850 of bf481.

Official start of PIcontrol DECK run is year 950.

Full history

  • u-ax871 from year 201 of at658 (JULES)
  • u-bd052 from year 301 of ax871
  • u-be106 from year 401 of bd052
  • u-bf481 from year 501 of be106
  • u-bi899 from year 850 of bf481
  • Official DECK start at 950 of u-bi889
Last edited 9 days ago by Martin Dix (previous) (diff)
Note: See TracTickets for help on using tickets.