CLM4.5 PTCLM error. Anyone have run with PTCLM success?

[sunshaobo133@163_com]

When I run my site data with PTCLM mode, there are some problems as follow:Creating data domain
Run command: /wps/home/shaobs/cesm1_2_1/tools/mapping/gen_domain_files/gen_domain -m /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapdata/map_1x1pt_QH-Arou_noocean_to_1x1pt_QH-Arou_nomask_aave_da_141203.nc -o navy -l 1x1pt_QH-Arou -c 'Running gen_domain from PTCLM' > /wps/home/shaobs/cesm1_2_1/scripts/gen_domain.log
ERROR(./PTCLM.py):: Error command does NOT exist: /wps/home/shaobs/cesm1_2_1/tools/mapping/gen_domain_files/gen_domain
Aborted (core dumped) If anyone have the same problem? help me Thank you

 

[jedwards]

You need to go to the directory /wps/home/shaobs/cesm1_2_1/tools/mapping/gen_domain_files/  and follow the instructions there to build the gen_domain tool. 

 

[sunshaobo133@163_com]

dear jedwards:       In order to test mine CESM porting, With the US-UMB site data I have follow the user's guide method as follow:> cd models/lnd/clm/tools/clm4_5/mksurfdata_map
> gmake
> gmake clean
> cd ../../../../../../tools/mapping/gen_domain_files/src
> ../../../../scripts/ccsm_utils/Machines/configure -mach sunpc -compiler intel
> gmake
> gmake clean

I find I can't do the "gmake" under the "mksurfdata_map" directory and must do this under the "../mksurfdata_map/src" directory. If these appropriate?

Then I do the configure (./../../../scripts/ccsm_utils/Machines/configure -mach sunpc -compiler intel) it success, and then do the gmake as follow it also sucess.

[shaobs@login04 src]$ gmake
ifort -c -DFORTRANUNDERSCORE -DNO_R16 -DFORTRANUNDERSCORE -DNO_R16 -DLINUX -DCPRINTEL -O2 -fp-model source -convert big_endian -assume byterecl -ftz -traceback -I/wps/home/shaobs/programFiles/netcdf4.3.2/include -I/wps/home/shaobs/programFiles/netcdf4.3.2/include gen_domain.F90
ifort -o ../gen_domain gen_domain.o -L/wps/home/shaobs/programFiles/netcdf4.3.2/lib -lnetcdff -lnetcdf -L/wps/home/shaobs/programFiles/lapack3.4.2 -llapack -lblas

Last do the PYTHON.py the errors are:

Re-create mapping files for surface dataset:
Run command: /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapdata/mkmapdata.sh --gridfile /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapgrids/SCRIPgrid_1x1pt_US-UMB_nomask_c141204.nc --res 1x1pt_US-UMB --gridtype regional > /wps/home/shaobs/cesm1_2_1/scripts/US-UMB_ICRUCLM45BGC_QIAN/mkmapdata.log
ERROR(./PTCLM.py):: Error running command: /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapdata/mkmapdata.sh --gridfile /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapgrids/SCRIPgrid_1x1pt_US-UMB_nomask_c141204.nc --res 1x1pt_US-UMB --gridtype regional > /wps/home/shaobs/cesm1_2_1/scripts/US-UMB_ICRUCLM45BGC_QIAN/mkmapdata.log
Aborted (core dumped)


I have check the mkmapdata.log file and it shows:

Using user specified scrip grid file: /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapgrids/SCRIPgrid_1x1pt_US-UMB_nomask_c141204.nc
Output grid resolution is 1x1pt_US-UMB
Machine login04 NOT recognized
Path to ESMF binary directory does NOT exist:
Set the environment variable: ESMFBIN_PATH

It shows the ESMF problem, how should I do these? I have do it in .bashrc but i doesn't work.


Thank you

 

[sunshaobo133@163_com]

I have set the ESMF env asexport ESMFBIN_PATH=/wps/home/shaobs/programFiles/esmf6.2/bin/binO/Linux.intel.64.mpiuni.defaultthe above problem was solved, but a new problm appears as:
Re-create mapping files for surface dataset:
Run command: /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapdata/mkmapdata.sh --gridfile /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapgrids/SCRIPgrid_1x1pt_US-UMB_nomask_c141204.nc --res 1x1pt_US-UMB --gridtype regional > /wps/home/shaobs/cesm1_2_1/scripts/US-UMB_ICRUCLM45BGC_QIAN/mkmapdata.log
rm: cannot remove `PET*.Log': No such file or directory
ERROR(./PTCLM.py):: Error running command: /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapdata/mkmapdata.sh --gridfile /wps/home/shaobs/cesm1_2_1/models/lnd/clm/tools/shared/mkmapgrids/SCRIPgrid_1x1pt_US-UMB_nomask_c141204.nc --res 1x1pt_US-UMB --gridtype regional > /wps/home/shaobs/cesm1_2_1/scripts/US-UMB_ICRUCLM45BGC_QIAN/mkmapdata.log
Aborted (core dumped) check the log file: ERROR: Problem on processor            0
 . Please see the PET*.RegridWeightGen.Log files for a traceback.
Output mapping file was NOT created: map_0.5x0.5_AVHRR_to_1x1pt_US-UMB_nomask_aave_da_c141204.nc It's the processor problem, where and how should I set these? Thank you 

 

[zhu467@purdue_edu]

Hi, shaobo,Have you solved this problem when using PTCLM? I am also stuck in this part.cheers.

 

[sunshaobo133@163_com]

Hi, zhu,     I have no idea about this problem now, but I think it may be resulted by the ESMF, you should check if you have install the ESMF successfully. I did't try this, now I just using the CLM4.0. Good luck!
Shaobo Sun

 

[cynthia_nevison@colorado_edu]

I tried to do a single point simulation for the first time, using -pts_lat and -pts_lon in my create_newcase command.  My problem/question is that the single point model didn't seem to run any faster than the full model would have.   It produced monthly output files at the rate of about 2.5/minute, even though those files were much smaller than normal history files.  I left the time at -W 2:00 in my run file, thinking I could run for a thousand years with only a single point, but instead the model only made it to 25 years before the 2 hours ran out.  Aside from changing COLD_START to ON, and changing the run time, I made no changes to env_run.xml or any of the other env_*xml files, and was running an out-of-the-box version of CESM1.2.0.  How fast can I expect PTCLM to run?  Do I need to change some other parameters to get it run faster?  Thank you

 

[zhu467@purdue_edu]

Since CLM can not run in parallel on a single point, it is quite often to cost 2h to run for 25 years.

 

[cynthia_nevison@colorado_edu]

If PTCLM only runs 25 years in 2 hours of wall clock time, and the run can't be continued from a restart file, how can PTCLM effectively be used to test code modifications that affect C and N pool sizes and thus need to be spun up for centuries? Thanks, Cynthia