Testing RSM with rsm script

If you use RSM for the first time, try using rsm scirpt for a test.




The forecast code runs 6 hours (model hours) at a time up to 48 hours. The model produces output files (r_pgb.ftxx) every 3 hours. See G-RSM scripts page for parameter definitions. The 48-hour simulation takes up to 1 hour (wall clock time) on a single processor, depending on the processor power.

If the model fails, error messages will be printed in r_fcstout.ftxx files. If the model fails prior to the forecast code, look at files such as rinpg.out, rmtn.out, scnvrt.out, and sfc0.out in order to find out what went wrong.

Files related to questions:
r_fcstout.ft2358 - Provided by Sai Ming



RSM blew up

hko 23 November 2006 03:57:22

I encountered an RSM blow-up (this is my guess) situation. Details of my setup are as follows:
Start date and hour - 1997 0501 00 UTC
Calculate six-hourly forecast up to 1997 0901 00 UTC
Initital data - sfc.1997050100.ieee, sig.1997050100.ieee
SST and ice gribs also from tsuzumi server.
I ran GSM first to generate the boundary conditions for RSM. No problem with GSM.
The r_fcstout.ft2358 file is attached on the front page.

Sai Ming

khideki 23 November 2006 06:25:14

Try decreasing the time step
in the rsim/csim script.
They are called DELTAT* and are
different for seasons.
Just run the model again. It will
restart from the last time restart files
were saved. Hideki


rsm script with SST update

lct 13 December 2006 01:55:31

I have a question about the SST update in rsm script.
In rsm($rsmcvs/runs/rsm), I can' see any SST update in forecast loop.
Should I modify it and put some script in rsm ?

khideki 13 December 2006 17:59:34

rsm script is for test purpose. After making sure rsm script works fine on your system, please use rsim script. rsim script uses SST files that have daily values. Hideki


rsm script - intel mpi

johnlow 29 April 2007 07:02:21

Dear Sir,

I am using intel compiler and mpich2 to test run the script. The error reported by rfcst.x is as follows:

Attempting to use an MPI routine before initializing MPICH

For your info, when I ran the grsmrt script with intel and mpich2, I was able to run for the gsmrt but not the rsmrt portion. The same error was reported. Appreciate your help.

John Low

0nmc regional spectral model-current version created march 2003
ncpus = 1
-1. 0. 0. 6. 6. 0. 6. 0. 3. 0.
360. 21600. 0. 1. 1. 0. 48. 0. 1. 24.
6. 3. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
rsm: run regional only with global files
fftfax is called in rftini. igrd2 = 108
in latbnd beta= 2.777777777777778E-004 factor= 15.0000000000000
------ initiate latbnd finished ----
------- into rsminp -----
emc ncep sigma surface file
lab=" emc ncep sigma surface file "
----- begin rsetsig ----
delx dely 60000.0000000000 60000.0000000000
Attempting to use an MPI routine before initializing MPICH

khideki 02 May 2007 01:50:49

Can you tell me the parameters you set in configure-model? NCPUS should be 1 and NPES should be the number of processors you want to use. The error does not seem to be at script level. There may be something wrong in the way the model was compiled. Also please try to find out if your system requires something to be done to "initialize mpich". Hideki


rsm script got "Segmentation fault"

yusengao 28 August 2007 04:13:29

Dear Sir,

When I test rsm script to run "./rsm",I got mesages as follow :

++ ln -fs r_slmsk fort.51
++ ln -fs r_orogrd fort.52
++ ln -fs r_mtnvar fort.53
++ ln -fs r_oroggd fort.54
++ ln -fs r_smooth_oroggd fort.55
++ rm rmtn.x
++ ln -fs /hptc_cluster/gg/work/zrsm/rsm/bin/rmtn.x rmtn.x
++ ./rmtn.x
/hptc_cluster/gg/work/zrsm/rruns/runscr/r_mtn: line 42: 28393 Segmentation fault ./$PROG.x <rmtn.parm >$PROG.out 2>&1
++ '[' 139 -ne 0 ']'
++ exit 8
+ exit 8

I didn't get output.
My mechines is HP_XC,LINUX X86_64 AMD CPU,compiler gcc,pgf90.
I don't change any thing.

yusengao 28 August 2007 05:27:56

Dear Sir,
This question is my machines problem.I already solved.
I run rsm in login node to be wrong.
I become to use "bsub -n1 -I ./rsm" to work to "compute node",looks like as if normally.


rsm script got "file not exist" error

yusengao 05 September 2007 05:35:24

Dear sir,

I run ./rsm to get error in ./sfc0.x .
And I change to ./r_000 directory to view sfc0.out ,
then got error messages

" PGFIO-F-209/unformatted read/unit=19/'OLD' specified for file which doe
s not exist.
In source file naopen.f, at line number 40 "

In F&Q ,to say "uncompress" command may be not exist in this problem.
But my machine have it.

I try to " vi naopen.f " to delete "access" like that

" c open(lu,file=cfn,iostat=iret,access='direct',recl=isize,
c 1 form='unformatted')
c read(lu,rec=1) bbuf
1 form='unformatted')
read(lu,rec=1) bbuf
,and re-compiler.

I view the "sfc0.out" ,that get error to become
" PGFIO-F-202/unformatted read/unit=19/conflicting specifiers.
File name = /hptc_cluster/gg/work/zrsm/zrsm-rsm-single/libs/con/clim.t
g3.grib unformatted, sequential access record = 0
In source file naopen.f, at line number 43 "
So I think file should be exist .
The problem perhaps with platform or compiler related.

My machine infomation is :
HP_XC 4000 Culuster AMD Opterron Linux X86_64
Red Hat Enterprise Linux version 4
compiler used : gcc , pgf90

yusengao 06 September 2007 07:49:07

This question solved.
The question is on the file 'nainit.f'.
I change ' status(12) ' to ' status(13) ' in this file.
After the change , ./rsm may normally finish.