Error running CAM5 standalone with data ocean

[junjie_liu@jpl_nasa_gov]

I tried to run CAM5 standalone version with data ocean, but it could not get started. The following is how I built and ran the model.

The model version is cesm1_0_3. I first ran the model with climate ocean, and has no problem. I used the following commands to build the model:

$cfgdir/configure -dyn fv -hgrid 1.9x2.5 -spmd -nosmp -debug -ntasks $procs -usr_src $camroot/models/new -fc ftn || echo "configure failed" && exit 1
echo "building CAM in $blddir ..."

$cfgdir and $camroot were defined. The executable were built without any problem.

Then I ran the model with climate ocean for several years with the following namelist:

$cfgdir/build-namelist -s -config $blddir/config_cache.xml -case $case -runtype $runtype
-namelist "&camexp start_ymd=00101, start_tod=00,
stop_option='nmonths' stop_n=$stop_n ,inithist='MONTHLY',nhtfrq=0,
ncdata='$CSMDATA/atm/cam/inic/fv/cami-mam3_0000-01-01_1.9x2.5_L30_c090306.nc',
/" || echo "build-namelist failed" && exit 1

At last, I wanted to continue the model run but with data ocean. I did not rebuild the model, since I thought they could use the same executable. But I changed the build-namlist commands:
$cfgdir/build-namelist -s -config $blddir/config_cache.xml -case $case -runtype $runtype
-namelist "&camexp start_ymd=00101, start_tod=00,orb_iyear=2000,sstcyc=.false.,
stream_year_first=1850,stream_year_last=2009,
stop_option='nmonths' stop_n=$stop_n ,inithist='MONTHLY',nhtfrq=0,
ncdata='$CSMDATA/atm/cam/inic/fv/cami-mam3_0000-01-01_1.9x2.5_L30_c090306.nc',
bndtvs='$CSMDATA/atm/cam/sst/sst_HadOIBl_bc_1.9x2.5_1850_2009_c101028.nc',
/" || echo "build-namelist failed" && exit 1


The main differences between these two build-nameists is "sstcyc", and "bndtvs".

When I tried to run the model, I got the following error message:

dtime_sync= 1800 dtime_clm= 1800 mod = 0
(lnd_init_mct) :Atmospheric input is from a prognostic model
(seq_mct_drv) : Initialize ocn component
(shr_file_setIO) file ocn_modelio.nml non existant
_pmii_daemon(SIGCHLD): [NID 00957] [c4-2c2s1n3] [Thu Jan 26 19:00:03 2012] PE 0 exit signal Segmentation fault
[NID 00957] 2012-01-26 11:00:03 Apid 5439924: initiated application termination


I would appreciate your comments on how to run CAM5 standalone version with data ocean, and what could be the source of this error.

Best Regards,

junjie

 

[eaton]

I have been able to reproduce this behavior. Unfortunately the data ocean model (docn) is not providing helpful error messages. It turns out that the message:

(shr_file_setIO) file ocn_modelio.nml non existant

is not an error message, but is just informational. The real error appears to be that when docn is looking for data it expects to have a valid date. The run you're doing started from a climatology run, and used the standard convention of starting a climatology run in year 0. But when switching to a run using historical data the year (which is now a single digit number if you've done several years of spin-up) is not valid.

The other thing you've done which is potentially a problem is to use the restart mechanism for a run that is technically not a restart. When restarting a run it is assumed that you are just continuing the exact same run which produced the restart files. The only thing that can be changed during a restart is the ending date. When it is desired to continue a run, but to change something about the run configuration (such as the SST data) then the correct technique is to do an initial run. CAM provides the functionality to write an initial file at the end of a run which may be used to start up a new run with a different model configuration. By default CAM writes an initial file at the beginning of each year.

I would suggest using an initial file from the end of the climatology run to start the subsequent run using historical SST data. In the new run use start_ymd to set the actual run date.

 

[junjie_liu@jpl_nasa_gov]

Brian,

Thank you very much for your help. It works now when I change the running mode to "startup" and provide the initial conditions.


Junjie

 

[parama_mukherjee@stonybrook_edu]

Hi,I am also facing the same problem in trying to use a historical (1950-2006) SST data set with CESM1.0.4. I get the same error : (shr_file_setIO) file ocn_modelio.nml non existantforrtl: severe (174): SIGSEGV, segmentation fault occurredI have changed the stream_year_first, last, sst file path name, sstcyc namelist variables in atm_in, docn.streat.txt etc and also my start_date is in the yyyymmdd format so not sure what I am missing out here. Any help would be greatly apreciated.Thanks in advance,-Parama 

 

[charan0239@gmail_com]

hii I am also facing the same problem, changed the sst file time axis to 1850-01-01 and regridded sst file to 96x144 (lat x lon - 1.9x2.5) with cdo, still getting same problem, could you able to solve the issue,please help me