[daip] Re: AIPS: SLFIT units

[Eric Greisen]

Jean Turner writes:

 > I have an AIPS question that has me stymied. I cannot figure out
 > the units when you do a plot to a SLICE plot using SLFIT.

      Welcome to the crowd - and I thought I had fixed it some.

 > 
 > When you do a Gaussian fit to a SLICE plot file - fitting the
 > guesses from the screen TKPL with TKSET - it fits a nice Gaussian to,
 > say, a spectral line, but then it gives answers in some unknown
 > units that do not look like pixels,  nor pixels*pixelincrement.
 > Y-axis is fine. X-axis has no units and this is the problem with
 > the FWHM and position of the fit. I realize that it fits the PLOT
 > and that the units will depend on LTYPE and the answers that
 > I get from SLFIT do indeed change (at least the 2nd set of values;
 > the first are in "SLPOINTS", I don't know what those are) when
 > I change LTYPE but I cannot for the life of me figure out what
 > units these are. When I plot pixels wrt to reference pixel (LTYPE=8?)
 > the answers sure do not seem to be pixels. If it were pixels from
 > reference I'd be golden, I can figure from there.

     I will need to see your IMHEADER, the input to SLICE, and maybe
EXTLIST on the SL file afterwards to help with details.  If your slice
is on a diagonal in disparate axes, then it is very hard to produce
any physical labeling.

     A slice is an interpolation at a bunch of points along a line in
the image - slice points are those interpolation points.  There are
256, 512, etc of them so they are not original image pixels but can be
related to them.  SLFIT gives two sets of answers - one in flux and
slice points and the other (not properly labeled in advance, sigh...)
in physical units if the physical units can be figured out.  My test
cases just showed a full RA axis (I did a horizontal slice) and arc
sec (a diagonal slice).

     Note that TKSET is unreliable.  Modern aips versions have a full
set of TV... verbs (TVSLICE, TVASL, TVSET etc etc) which are eliable.


 > 
 > In fact the output spectra seem to be coming out referenced to
 > the center of the SLICE box, if at all - this is not the same as
 > teh reference pixel.
 > 
 > (The FITS file is something that was not created in AIPS and the
 > CDELT1 was input using PUTH I believe, but it shows up okay with
 > IMH. Likewise reference pixel and value are fine in IMH)


Eric