Dave, Tim, Peter,

Many thanks for the responses, they are really useful.  I don't know  
the distortion as yet but will try autoastrom to define it.  The ZPN  
mapping sounds like the best bet, although defining a mapping then  
regridding to a uniform pixel scale might be a good idea so the images  
can be exported to fits.

cheers,
Colin

On Oct 22, 2008, at 10:30 PM, David Berry wrote:

> 2008/10/23 Peter W. Draper <[log in to unmask]>:
>> On Thu, 23 Oct 2008, Tim Jenness wrote:
>>
>>> Paul is right in that ZPN should work fine. AST supports that  
>>> without
>>> problem and you can set up polynomial distortions easily (we do  
>>> that with
>>> SCUBA-2). KAPPA REGRID can resample with a null mapping. You might  
>>> even find
>>> that autoastrom can calculate the distortion for you and attach a  
>>> reasonable
>>> WCS if you have enough stars (since ASTROM can handle this  
>>> distortion I
>>> think).
>>
>> That's right ASTROM will fit pincushion distortions, but the FITS  
>> output
>> from it uses a TAN projection with some non-standard polynomial  
>> terms, not
>> ZPN (we should investigate if that can also be done sometime). All  
>> the
>> KAPPA tasks understand this non-standard TAN projection.
>
> If you already know how the true pixel scale varies with radius,
> another option would be to add an extra mapping into the existing WCS
> to describe just the transformation from uncorrected to corrected
> pixel position. This avoids all messing about with RAs and Decs. (You
> can do this sort of thing with AST, but not with FITS-WCS because of
> the rigidly prescriptive nature of the FITS-WCS recipes. Instead of
> inventing a cookery book, they should have invented a standardised way
> of describing, communicating and implementing recipes, and left the
> writing of the recipes up to the observatories).
>
> To do this you would use the ATOOLS package. The ATOOLS:ASTMATHMAP
> command creates the correction mapping based on a set of user-supplied
> algebraic expressions giving corrected pixel positions in terms of
> uncorrected pixel positions. The ATOOLS:ASTREMAPFRAME command then
> adds this correction mapping into the WCS information in a specified
> NDF.
>
> I can supply more details about this approach if needed.
>
> The only point to note about this approach is that you will not be
> able to export the resulting WCS in FITS-WCS format. Not a problem if
> you will only be using starlink software. AST allows WCS recipes to be
> created arbitrarily from a wide range of basic ingredients, most of
> which do not appear in the very thin FITS-WCS cookery book.
>
> David