This broadcast is to announce a new version of the paper
describing the coordinates proposal by Mark Calabretta and myself.
The paper is available as wcs.all.ps, wcs.all.ps.Z and wcs.none.ps.Z
(only) from the fits ftp/WWW site (fits.cv.nrao.edu) in the
fits/documents/wcs directory.

There have been a lot of changes in the paper since the last broadcast
version of about two (!) years ago.  Some intermediate versions of the
paper were placed in the fits area, but not broadcast widely.  I have
attempted to run 'diff' to find out what the changes are, but expect
that I have missed some.  What I found is listed below:

1.  The paper has been converted to A & A format with suitable style
    changes, re-arranging of long equations, dropping excess figures,
    etc.

2,  The subject of default coordinate types (there is none now), and
    general longitude and latitude types is addressed.

3.  The finding of the native North pole from the reference point
    (when it is not the North pole itself) required a lengthy
    generalization and discussion and the introduction of the LATPOLE
    keyword.

4.  AIR projection mathematics corrected.

5.  CYL now called CYP.

6.  CEA definition of lambda corrected.

7.  Conical projections reference pixels now at the central Native
    latitude and PROJP1 gives the central latitude and PROJP2 gives
    1/2 of the difference of the 2 latitudes of the projection.  All
    conical projections required a little change, mostly in Y_0, in
    the math.

8.  COP - add an alternate form of Rtheta.

9.  COO - change symbols.

10. BON - simpler formulation.

11. GLS - correct formula for phi.

12. PAR - correct units.

13. QSC - better symbols, corrected formulae

14. TSC - extra form for x_f, y_f

15. Added appendix defining Pixel regularization image.

16. Added references to Calabretta's software package known as wcslib
    (version 2.0 now in preparation).

17.  Correct accuracy of numbers in numeric example.

18.  Generalize conics slightly to allow ref latitude < 0.

19.  Figures 30 and 31 had the computation of the difference between
     ZEA and SIN corrected.  ZEA is very similar to ARC and STG now.
     (Thanks to Steve Allen.)

20.  The translation between the PC matrix and the old CDELT/CROTA
     system was corrected.  The order of the matrix multiplies matters
     and was done wrongly in developing those formulae.

21.  References to the PROJPn, particularly in Table 5, were
     incomplete and out of date. 

22.  A strong recommendation to make the PC matrix positive as well as
     diagonal was added.  The sentences on CDmmmnnn and CDn_m were
     made into a separate (short) paragraph.

23.  The reference to wcslib was brought up to rev 2.3 and the
     misstatement about public domain was corrected.

24.  The increments in the pixel regularization image do not have to
     integer image pixels, but can be any appropriate value.  A
     statement to clarify this was added.

25.  The wording in the abstract re other kinds of coordinates was
     modified.

26.  An appendix by Bill Pence, Arnold Rots, and Lorella Angelini
     suggesting extensions to the main conventions to cover
     coordinates in tables has been included. 

27.  Casual wording regarding "reference pixel" was improved and
     generally changed to "reference point".

This seems like a lot - I guess #15 (and maybe #26) are the biggest
changes which should get wide discussion.  #2, 3, 7 are significant as
well, while #1 was surprisingly easy and mostly cosmetic.

Please let us know what you think.


Eric Greisen (egreisen@nrao.edu)
Mark Calabretta (mcalabre@atnf.csiro.au)