questions regarding CAM5.1.1 initialization

[murali@uni_no]

I am a beginner in CESM applications and facing some questions while running CAM-5.1.1 (standalone). I am setting this up for a simulation of the period 1980-2010 with observed, daily SSTs and ice fraction. After succesfully configuring and building the model, I am a bit confused regarding the proper initial conditions I should be using. Since I need time-varying forcing fields, I put 'sim_year=1850-2000' while building the namelist. If I leave 'start_ymd/ic_ymd' for defaults, this setup will take either 'cami_1980-01-01_0.47x0.63_L26_c071226.nc' or 'cami_0000-09-01_0.47x0.63_L26_c061106.nc' or 'cami-mam3_0000-01-01_0.47x0.63_L30_c100929.nc' depending on nlev. My questions are;

1) Are the latter two (cami_0000* and cami-mam3*) are climatological forcing files that should only be used for constant forcing, present day simulations? Additionally, why do they have different number of variables (the first two have around 30 and the cami-mam3* has 62)

2) If I want to start my simulation at, say, 1951, which initial condition file should I use?

3) In the default surface forcing file, the variables SST and ice_cov are prediddled so that the model takes correct values after time interpolation. How do I 'prediddle' fields in my observed surface forcing file?

4) With th efollowing build-namelist command, I could run CAM-5.1.1 successfully for a month starting from 1980-01-01:00:00:00 at 0.5 deg. resolution (with the default initial and surface forcings). The output looks reasonable. I just wonder if the changes I did to the defaults are correct.

$cfgdir/build-namelist -dir $rundir -config $blddir/config_cache.xml -case $case -config_cice $blddir/config_cache_cice.xml
-csmdata $inputdata -runtype $runtype -test -namelist "&camexp sim_year='1850-2000' start_ymd=19800101 start_tod=0
stop_option='ndays' stop_n=30 nhtfrq=-6 mfilt=5 hist_nhtfrq=-6 hist_mfilt=5 npr_yz=128,2,2,128 stream_year_first=1850 stream_year_last=2008
ncdata='$inputdata/atm/cam/inic/fv/cami_1980-01-01_0.47x0.63_L26_c071226.nc'
fpftdyn='$inputdata/lnd/clm2/surfdata/surfdata.pftdyn_0.47x0.63_hist_simyr1850-2005_c100826.nc'
fsurdat='$inputdata/lnd/clm2/surfdata/surfdata_0.47x0.63_simyr1850_c100826.nc' /" || echo "build-namelist failed" && exit 1

Any comments or suggestions ?

Thanks and regards,
Muralidhar

 

[eaton]

The choice of an initial file depends on the goals of the
simulation. Typically for climate simulations any initial file
the allows the model to run is OK, and one just discards the
first several months of simulation during which the atmospheric
state is spinning up to be in balance with the external forcings.
How long the spin up period needs to be is of course dependent on
the time scales of the features of interest, and how closely the
initial conditions are in balance with the external forcings.
The most important difference between the initial files you
mention is that the files containing 26 levels are for use with
the cam4 physics package while the 30 level file is for use with
the cam5 physics package.

In the default SST files the "prediddle" variables are the ones
that have only been spatially interpolated from the original 1
deg AMIP datasets. The variables that are used by the data ocean
component have had the additional time "diddling" applied as
described in detail here:
http://www-pcmdi.llnl.gov/projects/amip/AMIP2EXPDSN/BCS/amip2bcs.php

Since you are using daily rather than monthly mean data I suspect
that the time diddling is not needed.

Your build-namelist command indicates that you are using
the cam4 physics package (because the 26 level IC file is used).
It appears to be set up correctly to use a time varying SST
dataset. But there are other external forcings which by default
will be set to use present day climatological values rather than
time varying ones, for example the GHGs, aerosols, and ozone
forcings. To see how to set up a run where all the forcings vary
in time have a look at the use case file,
$cesm_root/models/atm/cam/bld/namelist_files/use_cases/1850-2005_cam4.xml.
You could also run the build-namelist command with the argument
"-use_case 1850-2005_cam4" and examine the resulting namelist files.

 

[murali@uni_no]

Thanks a lot, Eaton. This information was of a great help !