Incorrect wave directions at the boundary for a curvilinear grid

[GilesFearon]

Hello

I am trying to run WW3 release 6.07.1 on a curvilinear grid for the west coast of South Africa. I am using a grid and depths from an existing CROCO model, since I have it on hand and would like to couple in future. For boundary conditions I am creating my own spectral files from the CMEMS reanalysis.

I am fairly confident that the spectral boundary files are well set up in WW3 format but when I run the model, the wave directions entering the model at the boundaries are clearly incorrect (see example attached). My hunch is that the error has to do with the model not using the angle of the curvilinear grid correctly? I couldn't see anywhere where I can specify the grid angle as an input?

I tried running the model with differently formatted grid files and reading them bottom to top (IDLA=1) or top to bottom (IDLA=3) but always get the same result. FYI here are my grid files and namelist files.

Any assistance in this regard would be much appreciated!

Many thanks
Giles

[mickaelaccensi]

Hi Giles,

The first thing to check is the format of your spec file for the boundary.
Could you please try to run your CURV with spectra from our hindcast :
LOPS_WW3-GLOB-30M-E0170S315_201005_spec.nc
LOPS_WW3-GLOB-30M-E0170S330_201005_spec.nc
LOPS_WW3-GLOB-30M-E0170S345_201005_spec.nc

from
https://data-dataref.ifremer.fr/ww3/GLOBMULTI_ERA5_GLOBCUR_01/POINTS/2010/SPEC_SE/

Best,

Mickaël

[GilesFearon]

Hi Mickaël

Thanks for the reply!

I ran my domain with the files from your hindcast and it worked perfectly, so indeed it must be an issue with my spectral files. I am writing them using the to_ww3 function in the wavespectra library. I've had a close look at the data in the efth variable and it all looks reasonable to me when compared with your files (energy is in the appropriate direction and frequency bins), so I think there must be something else in the file which is throwing it off.

I'll continue to try and solve the issue, but in the meantime I've attached an example spec file in case you can see something obviously wrong with it.

Otherwise I'll just switch to using your hindcast for my boundaries! (I had thought of using CMEMS since we want to operationalise a coupled ww3-croco config in future and we could easily adopt the CMEMS forecasts if the hindcast works)

Thanks
Giles

cmems.lon17.00.lat-34.40.spec.zip

[GilesFearon]

I ended up getting it to work by re-ordering the direction dimension of my spectral files to match the example files you sent i.e. starting with a direction of 90 deg, and then moving counter-clockwise. I had energy in the correct bins, but had not considered that the ordering was important.

Many thanks again Mickaël for the example files, as that helped a lot!

Cheers
Giles

[mickaelaccensi]

you're welcome Giles !