F2010C5-CMIP6-LR

Owned by Qi Tang
Last updated: Aug 12, 2019 by Balwinder Singh
2 min read

Overview:

This is a special atmosphere-only compset with perpetual 2010 forcings based on CMIP6 input4mips data. The main purpose of this compset is for the HR overview paper to compare with the parallel HR compset.

Compset Creation Code:

place holder for github link

Ocean Initial Condition:

Land Initial Condition:

  • Use the year 2010 land restart file from the transient compset (F20TRC5-CMIP6) to maintain consistency between F2010 and AMIP simulations
  • File name: 20180716.DECKv1b_A3.ne30_oEC.edison.clm2.r.2010-01-01-00000.nc

Land Surface File:

  • File name: surfdata_ne30np4_simyr2010_c20181025.nc
  • Steps for generating this file can be found here.

Atmosphere Initial Condition:

  • Use the atmospheric state at year 2010 of DECK AMIP simulation
  • File name: 20180716.DECKv1b_A3.ne30_oEC.edison.cam.i.2010-01-01-00000.nc

Solar file:

Aerosol Emissions:

  • Use the same CMIP6 MAM4 aerosol emission files as F20TRC5-CMIP6, but specify the year to 2010.
  • DMS file: Identical to F2010C5-CMIP6-HR

Greenhouse Gas Concentrations:

Volcano Forcing:

Land Use/Land Cover:

Tuning:

  • Identical to A_WCYCL1850S_CMIP6


Running F2010C5-CMIP6-LR on a coarse (ne4) resolution:

Running this compset on a coarse grid can greatly help in debugging where we can run the model with minimal number of tasks with a fast turnaround time. Following is a naive attempt to run this compset on the ne4 grid on Compy using the following series of commands.:

git checkout maint-1.0
cd cime/scripts
./create_test SMS_D.ne4_ne4.F2010C5-CMIP6-LR --compiler intel -p e3sm -t 2010tst

We do expect the model to crash. Following is the list errors and their solutions:

  • " NC_FORMAT_NETCDF4 is not yet supported"
    • Change PIO_TYPENAME to netcdf in env_run.xml file to get past this error for the time being (thanks to Robert Jacob). Robert Jacob is working on converting files from netcdf4 format to netcdf 3 format
  • UrbanInputldomain and input file do not match dims
    UrbanInputldomain%ni,ni,= 866 48602
    UrbanInputldomain%nj,nj,= 1 1
    UrbanInputldomain%ns,ns,= 866 48602
    • Default was to use "/inputdata/lnd/clm2/surfdata_map/surfdata_ne30np4_simyr2010_c20181025.nc" for "fsurdat" in lnd_in. If we switch this file with "inputdata/lnd/clm2/surfdata_map/surfdata_ne4np4_simyr2000_c160614.nc'" ne4 file, the model gets past this error.
  •  /inputdata/lnd/clm2/initdata_map/20180716.DECKv1b_A3.ne30_oEC.edison.clm2.r.2010-01-01-00000.nc 1966080
    check_dim ERROR: mismatch of input dimension 16368 with expected value 407 for variable gridcell 
    ERROR in E3SM/components/clm/src/main/ncdio_pio.F90.in at line 359
    • The initial condition file for land is based on ne30 grid. We can get past this error by using land cold start (finidat='' in user_nl_clm)
  • ERROR: negative layer thickness. timestep or remap time too large
    • The model is still using atmospheric initial condition file ("ncdata") which is on the ne30 grid. Changing the initial condition file to a file on the ne4 grid fixes this issue ( "ncdata = 'inputdata/atm/cam/inic/homme/cami_mam3_Linoz_ne4np4_L72_c160909.nc'"  in user_nl_cam).


The model can run on ne4 grid but we need to supply it with the input files on native (ne4) grid.

Outstanding issues:

  • We need an atmospheric initial condition file on ne4 grid for this compset (if we are not okay with the current file)
  • We need to decide whether it is okay to run the model with a land cold start
  • We need to decided on the "fsurfdata" file for ne4 grid.