Opened 18 years ago

Last modified 18 years ago

#5 new defect

pb for ORCA025 grid initialisation : missing tmask ?

Reported by: pinsard Owned by: somebody
Priority: major Milestone: adaptation 1/4 de degré
Component: component1 Version:
Keywords: Cc:

Description

ORCA025 grid initialisation : missing tmask ?

Here is the command and the error message :

IDL >  initocemesh,'ORCA025', DRAKKAR_EXP='G42'
...
% Compiled module: NCDF_MESHLEC.
% Compiled module: NCDF_LISTDIMS.
% Compiled module: TESTVAR.
% Compiled module: NCDF_LISTVARS.
% NCDF_VARINQ: Variable "tmask" not found.
% NCDF_VARINQ: Variable Inquiry failed, -1 is not a valid variable id.
% Execution halted at: NCDF_MESHREAD     369
  /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/ncdf_meshread.pro
%                      NCDF_MESHLEC       27
  /home/rech/eee/reee522/SAXO_DIR/SRC/Obsolete/ncdf_meshlec.pro
%                      INITOCEMESH       210
  /home/rech/eee/reee522/incas/geomag/geomag_ws/initocemesh.pro
%                      $MAIN$          

It is true that there is no tmask variable in /u/rech/cli/rcli002/ORCA025/ORCA025-I/ORCA025-G42_mesh_hgr.nc
but do I really need it ?!

nb : there is no tmask also in the other "mesmash" file /u/rech/cli/rcli002/ORCA025/ORCA025-I/ORCA025-G42_mesh_zgr.nc

So either I can work without tmask or I have to ask for a new kind of file
to DRAKKAR team.

Change History (5)

comment:1 Changed 18 years ago by pinsard

I have to use smallmeshmask to combine mesh and mask informations of ORCA025
written in these 3 files
ORCA025-G42_byte_mask.nc, ORCA025-G42_mesh_hgr.nc and ORCA025-G42_mesh_zgr.nc
but unfortunately files termilogies is not the one expected by smallmeshmask :

IDL> meshdir='/u/rech/cli/rcli002/ORCA025/ORCA025-I/'
IDL> smallmeshmask, 'ORCA025-G42_',iodir= meshdir
% Compiled module: SMALLMESHMASK.
% Compiled module: ISAFILE.
% Compiled module: REPORT.
% Compiled module: ROUTINE_NAME.
% Compiled module: STR_SEP.
% L.80 /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro: meshmask file(s) not found...
% L.80 /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro: /u/rech/cli/rcli002/ORCA025/ORCA025-I//ORCA025-G42_ does not exist
% L.80 /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro: /u/rech/cli/rcli002/ORCA025/ORCA025-I//ORCA025-G42_.mesh_hgr.nc does not exist
% L.80 /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro: /u/rech/cli/rcli002/ORCA025/ORCA025-I//ORCA025-G42_.mesh_zgr.nc does not exist
% L.80 /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro: /u/rech/cli/rcli002/ORCA025/ORCA025-I//ORCA025-G42_.mask.nc does not exist
IDL> 

By the way, what does smallmeshmask want to read in ORCA025-G42_ ? I don't see anything about this file in the documentation of this procedure.

what is the best option ?

. rename the files

. ask for some change in smallmeshmask ?

comment:2 Changed 18 years ago by pinsard

Before making any change in smallmeshmask.pro, I tried to use it like it is

$ . ./geomag_profile.sh -d $(pwd)
$ export IDL_STARTUP=${HOME}/My_IDL/init.pro
$ ln -s /u/rech/cli/rcli002/ORCA025/ORCA025-I/ORCA025-G42_mesh_hgr.nc ${GEOMAG_ID}/ORCA025-G42_.mesh_hgr.nc
$ ln -s /u/rech/cli/rcli002/ORCA025/ORCA025-I/ORCA025-G42_mesh_zgr.nc ${GEOMAG_ID}/ORCA025-G42_.mesh_zgr.nc
$ ln -s /u/rech/cli/rcli002/ORCA025/ORCA025-I/ORCA025-G42_byte_mask.nc ${GEOMAG_ID}/ORCA025-G42_.mask.nc

IDL> meshdir=getenv('GEOMAG_ID')
IDL> smallmeshmask, 'ORCA025-G42_','ORCA025-G42_smallmeskmask.nc',iodir= meshdir
% NCDF_CONTROL: Attempt to take the file out of define mode (ENDEF) failed.
                (NC_ERROR=-31)
% Execution halted at: SMALLMESHMASK     168
  /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro
%                      $MAIN$ 

line 168 is

ncdf_control, cdfidout, /endef
IDL>exit         
% Error opening file.
  File: /home/rech/eee/reee522/.idl/rsi/rbuf/history_rhodes_53073140
  Disc quota exceeded
$  ncdump -h /home/rech/eee/reee522/geomag_d/ORCA025-G42_smallmeskmask.nc
...

This file seems to be a correct NetCDF file but I have to know if it contains correct data (I saw quickly only and where is the problem of ENDEF. I saw quickly only FloatInf values.

Any way, organisation of files must be modified because of quota limitation on Idris filesystems.

comment:3 Changed 18 years ago by pinsard


After modifying environnement variables in order to read and write on GAYA, the
problem is still not fix ...

IDL> meshdir=getenv('GEOMAG_ID')
IDL> smallmeshmask, 'ORCA025-G42_','ORCA025-G42_smallmeskmask.nc',iodir= meshdir
% Unable to allocate memory: to make array.
  Not enough space
% Execution halted at: NCDF_TRANSFER      27 /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro
%                      SMALLMESHMASK     199 /home/rech/eee/reee522/SAXO_DIR/SRC/Grid/smallmeshmask.pro
%                      $MAIN$          

Is it a permanent or a temporary problem ? What to check ?

comment:4 Changed 18 years ago by pinsard

C'est sans doute un pb permanent (256 Mo disponible en interactif sur rhodes).

Je ne sais pas calculer la mémoire dont a besoin chaque étape de l'application.

Je continue donc la mise au point de l'adaptation ORAC025 sur aedon en changeant mes
répertoires d'IO pour /usr/temp/${LOGNAME}/.

comment:5 Changed 18 years ago by pinsard

A new pb occured when I run on aedon :

IDL> $echo $GEOMAG_ID                                                           
/usr/temp/fplod/geomag_d/
IDL> meshdir=getenv('GEOMAG_ID')                                               
IDL> smallmeshmask, 'ORCA025-G42_','ORCA025-G42_smallmeskmask.nc',iodir= meshdir
% Compiled module: SMALLMESHMASK.
% Compiled module: ISAFILE.
% Loaded DLM: NCDF.
% Compiled module: NCDF_LISTDIMS.
% Compiled module: NCDF_LISTVARS.
% NCDF_VARPUT: Value array has incorrect dimensionality. (3 v. 2)
% Execution halted at: NCDF_TRANSFER      28
  /usr/home/fplod/SAXO_DIR/SRC/Grid/smallmeshmask.pro
%                      SMALLMESHMASK     199
  /usr/home/fplod/SAXO_DIR/SRC/Grid/smallmeshmask.pro
%                      $MAIN$          

Line 199 is

  IF (where(varlist_ps EQ 'e3t_ps'))[0] NE -1 THEN $
    ncdf_transfer, cdfid, cdfidout, 'e3t_ps'

Is that mean that I don't have the right files (none of them contains e3t_ps) ?

At least now, SAXO expert can do their own tests on LOCEAN system because files
are on /usr/temp/fplod/geomag_d/.

An other good point is that ORCA025-G42_smallmeskmask.nc does not contain anymore only NaN values, so may be I can try to progress in step 1 ...

Note: See TracTickets for help on using tickets.