G-RSM scripts


Please post questions about scripts in G-RSM system to the discussion tab of this page.



rsm script parameters (default values)




rsim script parameters





Horizontal diffusion


This is not done in scripts but I'll write it here for now.

Horizontal diffusion coefficient of heat and moisture is difuh in rsm/src/rsml/rdeldif.f
Horizontal diffusion coefficient of momentum is difum in rsm/src/rsml/rdeldif.f




List of commonly used G-RSM scripts


rsm script
rsim script
gsm script
csim script - part I
csim script - part II



How the specification of climatology, analysis SST, Sea ice and snow work both in GSM and RSM.


1. If climatology file is specified, the model runs with climatology.
2. If analysis file is specified, the model runs with analysis. If the date on analysis file is not within 5 days, the program stop execution.
3. If no file is given, SST and Sea Ice are persisted (use the same as initial condition). Snow will be predicted from initial value. The rule is that if the model has prediction equation for that field, it will be predicted, otherwise, the initial value is persisted.
4. We now have a new option of persisting SST initial anomaly. This is accomplished by assigning initial analysis SST to SST_GRB_T0_FOR_PERSISTENCY in the run script or FNOBSST0= in the fcst.parm (rfcst.parm).

Discussions

hko

G-RSM run scripts

hko 09 November 2006 07:44:02

Hi,
I configured the GSM and RSM to run at T62L28. In the GSM script runs/runscr/fcst, the cloud tuning table libs/con/tune.t42l18.amip.ewmrg is used/referred. Same in the RSM script runs/runscr/r_fcst. According to the naming convention, the table should be related to T42L18 models. There is another cloud tuning table libs/con/tune.t62l28.reanl.ewmrg, which seems to be pertinent to T62L28 models. Should I amend the scripts to use the latter tuning table or leave the scripts as they are?
Sai Ming

mkanamitsu
mkanamitsu 09 November 2006 20:40:31

The cloud tuning table is resolution independent and you should continue using the default one. The file name simply means that it was generated using T62 version of the morel, and again the table can be used on any resolution. Besides, if you are using SLINGO scheme (default), the table is not even used by the model.

hko

G-RSM run scripts

hko 09 November 2006 08:56:14

Hi,
Is DELTAT_REG the time step (unit in second) used in RSM integrations?
Where I can find the diffusion coefficient in RSM?
Sai Ming

khideki
khideki 09 November 2006 17:35:56

Please see the main page of G-RSM run scripts.

kei

Scripts stop.

kei 29 December 2006 07:51:59

Hi

I hit a quite strange situation. I'd be happy if anyone helps...

[Problem]
test gsm/rsm scripts fail in (r)fcst.
For single-run: suddenly stops before reading glacier data.
For parallel(mpi)-run: instantly stops with segmentation faults.

[In Detail]
[gsm-single]

==== start reading albdfrac ===
file size= 1362312
file /home/kyoshimura/junk1/libs/con/clim.yhalbedo.grib opened. unit=
11
climatology file.
this is annual mean climatology
input data has max=0.1000E+03 min=0.1000E+00 k= 1 nzero= 55466 nmis= 0
lat/lon grid
imax,jmax of grib= 360 180
inttyp= 0
rmaxmin of model grid arry:
gauout has max=0.1000E+03 min=0.1354E+00 k= 1 nzero= 14600 nmis= 0

==

start
[gsm-mpi]
0.6795465752E-05 0.2363319567E-04 0.2756866693E+03 0.4596517993E-02
0.6891973456E-05 0.2406624427E-04 0.2724325545E+03 0.3725399175E-02
0.6849890590E-05 0.2515147887E-04 0.2683344838E+03 0.2877888237ESegmen
tation fault
Segmentation fault
Segmentation fault
-02

Of course stacksize is set unlimited.

Strange thing is, the scripts were OK yesterday. On yesterday, when I tried a high resolution run (T248L28), my workstation crashed seriously (processes could not be killed). Then I rebooted, and the scripts didn't work.

I re-installed PGI and MPICH, but situation didn't change...

Thnx, Kei.

kei
kei 29 December 2006 16:49:49

Self reply.

MPI run successfully worked after reboot. Seems like something mysterious happened... (because of remained cash or something, stacksize was not good, maybe...)

Single run still does not work properly, but I quit investigation on this because I won't use single-runs...

Thanks!