[5.5.4+]: CF 1.9 and Geostationary projection

[rschmunk]

Versions impacted by the bug

v5.x

What went wrong?

The latest CF Convention release (1.10) and apparently the prior release (1.9) now specify that the x and y coordinate variables must have standard_name attributes with values "projection_x_angular_coordinate" and "projection_y_angular_coordinate" rather than the previously specified values of "projection_x_coordinate" and "projection_y_coordinate". This thoroughly messes up using NJ's geostationary projected grid handling.

At first look, it seems that adding a couple more definitions to ucar.nc2.constants.CF, and then modifying the if statements in ucar.nc2.dataset.conv.CF1Convention and ucar.nc2.internal.dataset.conv.CF1Convention that discover the projection x and y axes is sufficient to fix this. At least doing so with my local copy of NJ was sufficient to enable working with the grid and to make a plot.

Before I submit a pull request, though, is there anything else I should keep in mind?

ETA: This was brought to my attention by someone at EUMETSAT, who said his particular group is trying to bring their data products up to compliance with CF 1.9 (or later).

Code of Conduct

• I agree to follow the UCAR/Unidata Code of Conduct

[haileyajohnson]

Nothing else I know of to consider, @ethanrd can you think of anything?

[ethanrd]

Only thing is (probably clear already but just to be complete and explicit) both projection_[x|y]_angular_coordinate and projection_[x|y]_coordinate need to be recognized as old data will still have the non-angular_ version. And it probably should not depend on the CF version as it is only a deprecation.

[rschmunk]

@ethanrd, What I did was add the angular coordinate standard name as acceptable. No way was I taking out the old non-angular coordinate standard name.

You can take a look at the June 14 commit on rschmunk/netcdf-java to see what little I found seemed necessary.

[ethanrd]

@rschmunk - Yeah, I figured you were on it. Looks good. Thanks!

[rschmunk]

Resolved by PR #1202.