save cube in TransverseMercator to grib2

[wideblue]

When trying to save cube in Transverse Mercator coordinat system into grib2 I get the following error

Traceback (most recent call last): File "/home/damjan/testGribWrite.py", line 46, in <module> iris.fileformats.grib.save_grib2(cubes[0],'/home/damjan/Namizje/apr1km.grib2') File "/usr/local/lib/python2.7/dist-packages/Iris-1.5.1-py2.7-linux-i686.egg/iris/fileformats/grib/__init__.py", line 706, in save_grib2 raise TranslationError("Did not find one (and only one) " iris.exceptions.TranslationError: Did not find one (and only one) latitude or longitude coord

It seems that save_grib2 expects lat/long coordinates and doesn't realise that it should save Transverse Mercator coordinat system. I'm not completely sure if this is an issue or saving a cube in TransverseMercator to grib2 is not supported or I have to set something up before saving.

[ajdawson]

I think currently our GRIB2 save support is quite limited, so unfortunately this is a case of Iris not supporting the operation rather than you doing anything wrong.

[bjlittle]

We've scheduled to have an iris-grib development sprint quite soon, and this is one of the use cases that we know we need to address.

Ping @dkillick and @pp-mo, is this still a live issue? I thought it was ...

[wideblue]

Ok, I saw no activiti on this and tought it would be best to close it. It's good to know there are plans for this issue.

[pp-mo]

this is one of the use cases that we know we need to address.

We have certainly had a request to read data on generalised Mercator grids.
I'm not sure we have been for transverse Mercator yet.
Also this request is to save, not to read it.

At present we support a good deal more for reading than saving, but there is nothing we can write that we can't read back in.
So I think we would want also want to add reading to this requirement.

It has occurred to me to suggest, as a principle, that we could move to supporting the same template types for writing and reading uniformly.
That would be simpler to explain, and could make testing easier.
O.K. I just proposed it : SciTools/iris-grib#98

[DPeterK]

We now have support for saving to Transverse Mercator in iris-grib. This was actually added in Iris at #1477 (before we split the GRIB handling code out into its own library).

@wideblue would you mind checking if this now does indeed work for you? It would be good to tidy up this issue. Thanks!

[pp-mo]

No reply, so I guess this is done.