Tools (#2) - [SIREN] Periodicity (#2) - Message List
I encountered an error when running create_bathy.exe :
FATAL : GRID GET INFO: can not read or compute domain periodicity from mpp bathy_coarse.nc. you have to inform periodicity in namelist.
The coarse namelist contains : in_perio0=0 , but this relates to the coordinates file, not the bathymetry file.
The problem is actually that Bathymetry (from the bathymetry.nc file) has no periodicity defined. I don't know how to specify that.
In the end, I just added 2 (hereunder, bold underlined) lines of code in grid.f90, subroutine grid__get_info_mpp :
IF( td_mpp%i_perio >= 0 .AND. td_mpp%i_perio <= 6 )THEN il_perio=td_mpp%i_perio ELSE ! look for attribute in mpp files il_attid=att_get_index(tl_mpp%t_proc(1)%t_att(:),'periodicity') IF( il_attid /= 0 )THEN il_perio=INT(tl_mpp%t_proc(1)%t_att(il_attid)%d_value(1),i4) __''''' ELSE'''''__ __''''' il_perio=0'''''__ ENDIF ENDIF
I suppose I did something wrong, but I couldn't find out what. With these 2 lines added to the code, it now works.
just to be clear, when I talk about ghost cell, I only think of the first point of the grid (for each boundaries), not of a buffer zone.
It should work on your Med configuration, as you "force" periodicity of the grid on your namelist (jperio).jpaul2016-04-14 09:56 CEST (5 years ago)
Thank you for your reply!
Actually I had already tried with bathy_coarse.nc containing only the variables Bathymetry, nav_lon and nav_lat (without longitude and latitude), this did not help.
bathy_coarse.nc, this is the file that I use for NEMO (when I run the parent grid only). Thi is actually a model of the Med, so there are a few columns of points west of Gibraltar. I did not put them to zero (even in the ghost zone). Maybe this is wrong, but NEMO seemed to accept that. I could run the model for a year without problems.
If it helps, I could put the first column (western boundary) to zeros, and/or add the global attribute. This way I don't have to go into the code next time
Lucluc2016-04-13 19:40 CEST (5 years ago)
SIREN should try to compute periodicity of your bathymetry file. To do so it uses ghost cell at boundaries.
For what I see from your bathy_coarse.nc, SIREN can't get the right one, however it should compute a periodicity.
So first you should remove from your file the latitude and longitude variable and keep only nav_lon and nav_lat. I think this is why SIREN do not succeed to compute periodicity.
Then as your domain is a regional domain, NEMO expected the first row on the western boundary to be a ghost cell. For bathy_coarse.nc it's mean you should force the firsts points of the western boundary to be 0. Doing so SIREN will be able to get the periodicity of your domain.
Another way to do it, should be to add a global attribute in_periodicity=0 to your file.jpaul2016-04-13 18:14 CEST (5 years ago)