Problem running CAM5 on 128x256 eul resolution

[sirajkhan78@gmail_com]

Hi,

I have run CAM5 for climatology and with prescribed SST with 64x128 resolution for euler spectral resolution. But when I try to run the same model for 128x256 euler resolution I getting error in build name list (Model configure and building works fine.) here is the error I am getting


Died at /home/siraj/cesm1_0/models/atm/cam/bld/build-namelist line 2572.
CAM build-namelist - ERROR: No default value found for ncdata
user defined attributes:
key=ic_md val=101

Can you help me to find out this problem
My script is copied as below.

#! /bin/tcsh -f

setenv P4_GLOBMEMSIZE 500000000
set netcdf = /home/siraj/netcdf
set mpich = /usr/local/mpi2
set cfg_string = "-fc ifort "
setenv INC_NETCDF ${netcdf}/include
setenv LIB_NETCDF /home/siraj/netcdf/lib
setenv INC_MPI ${mpich}/include/intel
setenv LIB_MPI ${mpich}/lib64

limit stacksize unlimited
setenv OMP_NUM_THREADS 1
set camroot = /home/siraj/cesm1_0
setenv CSMDATA /home/siraj/ccsm4_working_copy/inputdata

set cfg_string = "-fc ifort "
set version = "cam5"
set dycore = "eul"
set caseid = clim-cam5
set hgrid = "128x256"
set ntask = 32
set case = ${dycore}${hgrid}_${caseid}

set wrkdir = /home/siraj/cam
set blddir = $wrkdir/$case/bld
set rundir = $wrkdir/$case
set cfgdir = $camroot/models/atm/cam/bld
set archdir = $camroot/scripts/ccsm_utils/Tools/archiving

setenv LID "`date +%y%m%d-%H%M%S`"

mkdir -p $rundir || echo "cannot create $rundir" && exit 1
mkdir -p $blddir || echo "cannot create $blddir" && exit 1

if ( ! -x $blddir/cam ) then
cd $blddir || echo "cd $blddir failed" && exit 1
$cfgdir/configure $cfg_string -test
-phys $version
-dyn $dycore
-hgrid $hgrid
-chem none
-nlev 26
-ocn dom
-spmd
-nosmp
-linker mpif90
-ntasks $ntask

#|| echo "configure failed" && exit 1

echo "building CAM in $blddir ..."
rm -f Depends
gmake -j 4 >&! MAKE.out || echo "CAM build failed: see blddir/MAKE.out" && exit 1
endif

## Create the namelist
cd $blddir || echo " cd $blddir failed " && exit
touch namelistout.txt
echo " building the namelist..."
$cfgdir/build-namelist -case $case -test -config $blddir/config_cache.xml
-ignore_ic_year
-namelist "&exp start_type='startup'
stop_option='nyears' stop_n=22 restart_option='monthly' /"
>&! namelistout.txt
|| echo " build namelist failed - see $blddir/namelistout.txt for details " && exit
echo "build-namelist done."
echo " see $blddir/namelistout.txt to check the inputdata "





Regards
Siraj

 

[eaton]

build-namelist is telling you that it can't find a default initial file for the date 101. The file that contains the defaults is models/atm/cam/bld/namelist_files/namelist_defaults_cam.xml. Looking in that file you'll see that the only initial file for T85 has the attribute ic_ymd="901", in other words, it's an initial file for Sept 01. Since you're just interested in producing a climatology then the easiest workaround is to set start_ymd=901 in your build-namelist command. In CAM3 and earlier when the default dycore was Eulerian we used to use 901 as the default start date. But since CAM4 we've made the standalone CAM behavior more consistent with CCSM/CESM, and hence the switch to 101 as the default start date.

If you want to start your climatology run at 101 there are a couple of options. One would be to just do a 4-month run using the existing initial file, and generate a jan 01 initial file at the end of that run (it is the default behavior to write initial files at jan 01 of each year). You can then use that initial file in a new run that you start from 101. Another possibility is to just try using the 901 initial file for a run starting from 101. If the initial conditions aren't too out of balance the run might start this way (sometimes this doesn't work because a convergence criteria somewhere in the model won't be met). The easiest way to implement this option is to change the -ignore_ic_year option that you're currently using to -ignore_ic_date. That tells build-namelist to ignore the ic_ymd attribute when it is looking for defaults.

In looking at your configure command I've just realized that it is requesting an unsupported, untested, and probably unworkable configuration, i.e.,

-phys cam5 -nlev 26 -chem none

The cam5 physics package requires 30 levels (this is for the additional levels that have been added in the PBL) and it requires the modal aerosol package (-chem trop_mam3 which is the default when -phys cam5 is specified).

 

[sirajkhan78@gmail_com]

Hi Eaton,

Thanks for your continuous support. Today I started my CAM4 and 5 T85 resolution simulations. I have changed some default setting in models/atm/cam/bld/namelist_files/namelist_defaults_cam.xml file. I have used date 101 for the 128x256 resolution. After this my problem was solved but i get another error about problem i.e.

/hpc/home/sislam/ccsm4_working_copy/inputdata/lnd/clm2/surfdata/surfdata_128x256__c090213.nc
UrbanInput
/hpc/home/sislam/ccsm4_working_copy/inputdata/lnd/clm2/surfdata/surfdata_128x256__c090213.nc
Expected dimensions: lsmlon= 256 lsmlat= 128 netcdf error from check_ret:UrbanInput:NetCDF: Invalid dimension ID or name
ENDRUN: called without a message string

I again open the name list default file and have changed the surfdata_128x256__c090213.nc with surfdata_128x256_simyr2000_c100406.nc and then the problem was solved and my simulation is running.
I did these to remove my errors but don't know the physical reason about why runs for 128X256 need these changes in default name list.

And the last question is about the -phys cam5 -nlev 26 -chem none setting. Well for the comparison with CAM4 i have set nlevs to 26 in CAM5. I have completed the simulation using this setting. Do i need to start it again with chem setting ?. Does this will make much difference in simulation.

Sorry for many question.

Siraj

 

[eaton]

The cam5 physics was developed using the FV dycore (as was the cam4 physics). So I'm not surprised that things aren't going smoothly trying to use the Eulerian dycore. The reason we switched to the FV dycore is because it has better tracer transport conservation properties and this become increasingly important as more prognostic chemistry is incorporated into the model.

"-phys cam5 -nlev 26 -chem none" is a significant modification to the cam5 physics package. Both 30 levels and the modal aerosol code are part of the cam5 physics package. If you change either of those features you are not running the configuration that will be referred to as the cam5 physics package in the published literature, and can't claim to be comparing cam4 with cam5. CAM's configure utility allows you to make the changes you have made because CAM is a research tool. It allows you to produce many configurations that will not result in reasonable climate simulations.