Does Opendrift support ROMS grids without equidistant spacing?

[brucelaughlin]

Hello,

I ran example_roms_native.py, and noticed the following line being printed to the logger:

WARNING opendrift.readers.basereader.structured:51: No proj string or projection could be derived, using 'fakeproj'. This assumes that the variables are structured and gridded approximately equidistantly on the surface (i.e. in meters). This must be guaranteed by the user. You can get rid of this warning
by supplying a valid projection to the reader.

I also noticed that the grid used in example_roms_native.py has approximately even grid spacing on the surface.

I'm now wondering - does Opendrift currently support ROMS grids which are not gridded approximately equidistantly on the surface?

I have made a few attempts to supply a projection when using reader_ROMS_native.py, but I have not been successful.

Thank you

[knutfrode]

Yes, several users have been worried about this warning, so I now made it less scary:
79109cf

And yes, the ROMS reader works well also for curvilinear fields.

It is not possible to provide a projection to the ROMS native reader.
If data/file is projected, one can instead use the generic netCDF-reader, except if vertical coordinate is sigma, which is not yet supported by the generic reader.

[brucelaughlin]

Hi,

Thanks again for replying. I already replied to your other response to my other post, but perhaps I can help clarify the problem here.

In the warning that you changed, it used to say:

"... This assumes that the variables are structured and gridded approximately equidistantly on the surface (i.e. in meters). This must be guaranteed by the user"

So, I believe that the issue isn't with handling curvilinear grids. Instead, it's with handling grids which do not have equidistant spacing.

When I look at your example with the curvilinear Nordic ROMS grid, my analysis shows that you actually do have approximately equidistant spacing in meters between grid vertices across the entire grid, in both horizontal dimensions.

For the ROMS grid that I'm using, I have approximately equidistant spacing in degrees of Lat/Lon across the grid, but NOT in meters (it can't be both!).

THIS is what I believe leads to the strange offsetting that I tried to highlight in my other post. And it's why I suspect that Opendrift may not be accessing the correct velocities for interpolation, at least in some locations, when ROMS grids do not have equidistant spacing in meters.

Furthermore, my ROMS data does use sigma as the vertical coordinate, so I guess the generic reader won't work for me either!

I may be totally wrong, and I apologize if that's the case!

Cheers,
Bruce