Problems with SWAN in the COAWST system

[lvazquez]

Hi, I am trying to use a coupling between the ROMS-SWAN models, in the COAWST System. Is my first time using the SWAN model and I'm getting an error that I do not understand.

When I try to run the SWAN model the error obtained is:

Error : invalid boundary point
Error : ( -99 -99) is outside computational grid

Error : It is not recommended to use quadruplets
Error : in combination with zero wind conditions.

Any idea? I don't know where is the problem.

Thanks.

Liset

[jcwarner]

this looks like 2 different errors.

1) Error : invalid boundary point
Error : ( -99 -99) is outside computational grid

so one of your boundary points (maybe you are using tpar files?) is not in the grid. SWAN thinks the point is at -99 -99, which is south of the south pole. So maybe just comment out those boundary lines for now, get it to run, then slowly add the boundary lines back in until you get the line that causes the error.

2) Error : It is not recommended to use quadruplets
Error : in combination with zero wind conditions.
So you have selected to use the physics option 'QUAD', see swan manual for definition.
Either do not use this option, or, add winds. You can add winds with a constant wind such as
WIND 10. 25.
so this would be winds of 10 m/s towards 25 deg (north east)
or create a wind ascii file.

-john

[lvazquez]

Hi,

I solved the problem of the physics option 'QUAD', and I'm working on the error with the boundary points. I am using the TPAR files, I'm using as an example the Sandy Project that comes in the COAWST.

When I comment out the boundary lines I get the error
At line 1379 of file swn_outnc.f90
Fortran runtime error: Attempting to allocate already allocated variable 'recordaxe'

Maybe a problem with my fortran compiler??

If I use the boundary lines my problem are in the next lines:

BOUNDSPEC SEGMENT XY -91.2878 23.2430 -91.2929 23.3547 VARIABLE FILE 0 'Projects/Nate/TPAR18.txt'
** Error : invalid boundary point
segment point -91.29 23.35 grid -99.00 -99.00 -98 -98
** Error : ( -99 -99) is outside computational grid

BOUNDSPEC SEGMENT XY -79.6543 19.4963 -79.6645 19.7196 VARIABLE FILE 0 'Projects/Nate/TPAR19.txt'
** Error : invalid boundary point
segment point -79.65 19.50 grid -99.00 -99.00 -98 -98
** Error : ( -99 -99) is outside computational grid

BOUNDSPEC SEGMENT XY -79.7465 21.5066 -79.7567 21.7300 VARIABLE FILE 0 'Projects/Nate/TPAR28.txt'
point with location -79.7567 21.7300 is not active
** Error : invalid boundary point
segment point -79.76 21.73 grid -99.00 -99.00 -98 -98
** Error : ( -99 -99) is outside computational grid

BOUNDSPEC SEGMENT XY -90.8605 19.4468 -90.6064 19.4480 VARIABLE FILE 0 'Projects/Nate/TPAR33.txt'
point with location -90.6064 19.4480 is not active
** Error : invalid boundary point
segment point -90.61 19.45 grid -99.00 -99.00 -98 -98
** Error : ( -99 -99) is outside computational grid

BOUNDSPEC SEGMENT XY -91.2949 23.3537 -91.0408 23.3549 VARIABLE FILE 0 'Projects/Nate/TPAR65.txt'
** Error : invalid boundary point
segment point -91.29 23.35 grid -99.00 -99.00 -98 -98
** Error : ( -99 -99) is outside computational grid

I do not understand why it takes the point in -99, -99 in those boundary segments. I guess it's an error in the swan INPUT file when defining the grid, but I do not know.

I attach the INPUT file that I am using to create the init file for swan, running SWAN in stationary mode.

Any idea?

Thanks, Liset

[jcwarner]

ok.. 2 things here.

1) Comment the Boundary lines, and then you get the 'recordaxe' error. Comment out the writing of netcdf files for swan. I have a problem when both the parent and child are writing out netcdf files. I need to correct this but just did not get to it yet. You can write out mat files for both parent and child, just not netcdf files.

2) You seem to have 63 TPAR files, so i guess you created these. It is telling you which files are bad: 18 19 28 33 63. So comment out those TPAR file lines and see if it runs.

[lvazquez]

Hi, first of all, thanks for your help.

I comment the boundary lines and the writing of netcdf files for swan and working.

Now I am trying to do a coupling between ROMS and SWAN models, using 3 ROMS domains and 1 in SWAN. First I am creating the file of weights using scrip_coawst, but I have this error:

Reached the end of mapping one set of grids
end of scrip package
=============================================
Begin mapping between the two grids
---------------------------------------------
The interp file is: ocn3_to_wav1_weights.nc
The src grid is: ../../Projects/Nate/Nate_roms_grid_ref1.nc
The dst grid is: ../../Projects/Nate/Nate_swan_coord_ref3.grd
----------------------------------------------
grid1 dimensions= 617 167
----------------------------------------------
----------------------------------------------
grid2 dimensions= 452 176
----------------------------------------------
Using latitude bins to restrict search.
Computing remappings between: ../../Projects/Nate/Nate_roms_grid_ref1.nc
and ../../Projects/Nate/Nate_swan_coord_ref3.grd
grid1 sweep
grid2 sweep
STOP integration stalled: num_subseg exceeded limit

Why is that?? Is a memory problem o something like that? is due to the high resolution used in those domains?

Any idea??

I attach the scrip_coawst.in file that I am using.

Thanks, Liset

[jcwarner]

do me a favor and lets post this issue here:
https://coawstmodel-trac.sourcerepo.com ... /newticket

[lvazquez]

Hi,

I posted the issue there.

It is the Ticket #104. I hope that someone can help me with that problem.

Thanks, Liset