[fitswcs] Re: distorted TAN projections for SIRTF

[Mark Calabretta]

On Mon 2000/11/20 17:29:59 -0800, David Shupe wrote
in a message to: mcalabre at atnf.csiro.au

>Do you feel that we should avoid the IRAF TNX projection?  If so,
>will the PVi_xs representation of a distorted gnomic projection be 
>implemented in WCSLIB soon?

Dear David,

When I announced the last version of WCSLIB (2.5) I stated that the next
version would be released once the proposal is ratified and that it would
handle the extended range of parameters required by the TAN distortion
function.  In fact, I envisage that it will handle everything in papers 1
and 2, including the pixel regularization correction.

However, I am reluctant to make any changes to WCSLIB, particularly ones
which will affect the API, until I have some measure of certainty that the
proposal will be adopted in close to its current form.  When I announced
version 2.5 last December (see appended) I was thinking, wishfully it
seems, in terms of a timescale of several months.

Since then there has been little progress towards formal ratification,
there is not even an "in principle" agreement in any formal sense, though
informally I don't envisage that there will be any major changes.  There
has been some background discussion but the main result has been the
addition of more explanatory material to the celestial coordinates paper;
the technical details themselves remain essentially unchanged from the
mid-1999 draft.

Nevertheless, the stated proviso that the proposal be ratified before
WCSLIB 3.0 is released is still the determining factor and it would appear
that in referring your question to me the FITS BoF has overlooked this
fact.  In the meantime, WCSLIB 2.5 provides ample "proof of concept" for
the celestial WCS proposals.

Consequently, I've CC'd this response to the email list recently set up
to debate the adoption of the WCS proposal.  This group is essentially in
control of the process and should be in a better position to answer your
question.

Regards,
Mark Calabretta
ATNF

>>>

... once the FITS WCS proposal is accepted by
the IAU I envisage that the next release of WCSLIB (3.0) will implement

   1) The change from CDELT+PC matrix to CD matrix.

   2) Enhanced range of projection parameters, from the present 10 (total)
      to 100 on each axis (i.e. 200) to handle astrometric plate solutions
      in the TAN projection.

The first of these will require a change to the interfaces of the C and 
FORTRAN libraries, the second will only affect the FORTRAN library.

>>>

On Mon 2000/11/20 17:29:59 -0800, David Shupe wrote
in a message to: mcalabre at atnf.csiro.au

>Dear Dr. Calabretta,
>
>I would like to ask your help regarding FITS representations of
>distorted images.
>
>Several of us at the SIRTF Science Center are defining data products
>for the SIRTF instruments.  The imaging instruments have distortions
>of a percent or more.  As our Basic Calibrated Data products will not
>include regridded data, we need to represent the distortions in the
>WCS included in the FITS headers.
>
>We are considering whether to use the "TNX" projection that is included
>in IRAF, or to use the PVi_xs keywords in the proposed FITS standard.
>Several scientists would like to use the TNX projection because of
>its availability in IRAF and other programs such as SAOimage.
>However, at last week's ADASS meeting, Bob Narron from our group
>put this question to participants of a FITS session (including Doug
>Tody and other IRAF representatives).  The recommendation of this
>group was not to use the TNX projection, but instead to use the
>PVi_xs keywords.  This group further recommended that we contact you
>to ask whether the PVi_xs keywords would be implemented in WCSLIB in
>the near future.
>
>Do you feel that we should avoid the IRAF TNX projection?  If so,
>will the PVi_xs representation of a distorted gnomic projection be
>implemented in WCSLIB soon?
>
>For the IRAC instrument on board SIRTF, a second-order polynomial
>is sufficient to represent the distortion.  We are still assessing
>what is needed for the other imaging instruments.
>
>Regards,
>-----------------------------------------------
>David Shupe               SIRTF Science Center
>Work:(626) 395-2357       MS 314-6
>shupe at ipac.caltech.edu    Caltech
>                          Pasadena, CA 91125