[fitsbits] polar coordinates

[David Berry]

On 11 December 2013 13:19, Walter Jaffe <jaffe at strw.leidenuniv.nl> wrote:
> I find nothing in the FITS standard document about representing the sky
> in polar coordinates.  One of my clients wishes to represent a
> theoretical model of disk emission in polar logarithmic coordinates.
> In other words CTYPE1 is LOG(RADIUS) (common log of
> radius in degrees on the sky) and
> CTYPE2 is PA (or POSITION ANGLE), (degrees E from North).
>
> Has this been discussed before?  Is there a straightforward
> convention?

I don't think you can do this in FITS-WCS. Firstly, I'm not aware of
any way of specifying an axis as co-latitude rather than latitude, and
secondly I'm not aware of any way of specifying an axis as log of a
projected sky axis (maybe paper IV may have covered this but we do not
yet have a published paper IV).

Maybe this is a good example of why we need a more flexible WCS system...

For what it's worth, you could do what you want using the AST library.
But obviously, when exporting the data you would need to use
AST-native encoding of the WCS rather than standard FITS-WCS encoding,
so this would restrict the applications that could use the WCS.

David