Re: New FITS Header

[Tom Droege <tdroege@veriomail.com>]

Jure and others for comment,

Thanks for all your efforts, I am just trying to get this straightened out 
without understanding everything.

I really need to know what the various items should be for what I am 
writing.

Maybe also include

BZERO   =              32768.0
BSCALE  =                  1.0

These are rejected by CFITSIO.  These are passed by Rob's program to 
CFITSIO and it rejects them saying that they are reserved words.

I am really confused.  for north to the right, what should
CTYPE1 and CTYPE2 be?

I don't think I will change this orientation at this time.  This is the way 
they come off the camera, and transposing them would take computation time 
which would slow down every exposure.

I am sending this to the list for comments by others.  Is it worth slowing 
down the data rate to have the images rotated?  How much computation time 
would this actually take?

Tom Droege


Tom

It does make the things easier.  For example, the cfitsio library reads all of
the specified keywords using a single command.  The items that you specify 
define
the coordinate system and give the initial approximation which is later refined
to the maximum precision using the star matching software.  If any of the
keywords is missing, the analysis program in principle doesn't know how to
calculate the coordinates of the pixels.  So, please specify everything.  There
is no need for any particular order, but for the ease of reading you may 
specify
information for one axis together, like

CTYPE1  = 'RA---TAN'
CRVAL1  =   1.636931844239E+02
CRPIX1  =   1.021500000000E+03
CDELT1  =   2.150153665884E-03
CROTA1  =   9.012538041574E+01
CROTA2  =   9.012538041574E+01


and equivalently for the other axis.  CTYPE1 and CTYPE2 define the type of
projection.  Instead of TAN there could be something else, but for the optical
telescopes it usually isn't.  CDELT1 and CDELT2 are the increments of the
celestial coordinates at the reference pixel position when you advance for one
pixel.  Some people will say this is the pixel size (expressed in decimal
degrees).  However, it may have a negative value.

Also note that in reality you have CTYPE1 = 'DEC--TAN'.  This would then give
CROTA1 = 0 (it is 90 degrees now).  But, for me (and maybe some other 
people, but
not all) it is easier to look at images which have RA changing in the 
horizontal
direction, what happens in real worild when you stand outside and look south.
This is why I recommended in the TN67 to switch the axis either by rotating the
camera or transposing the array in memory (you also need to adjust NAXIS[12] if
you do this) before you write out.



				regards, Jure