On Mon, 19 Apr 2004, David Berry wrote:
> Tim,
>
> > > Note the IMAGFREQ support for DSB instruments.
> >
> > When we switch to NDF writing in ACSIS we will have to have two specframes
> > in dual sideband instruments. I don't think IMAGFREQ support will be
> > possible in the AST encoding unless SpecFrames themselves are extended to
> > support dual sideband frames (there would need to be two or three new
> > attributes, one to indicate which sideband is current (and which was the
> > primary sideband), the other to indicate the IF. Switching to the image
> > sideband essentially means adding or subtracting the IF and multiplying
> > CDELT by -1. Maybe there could be a DSBSpecFrame which inherits from
> > SpecFrame).
>
> The best way to do this depends on what you want to do with the WCS. As an
> example, you have two DSB data sets, A and B. In A, the current WCS
> Frame represents the lower sideband, and in B it represents the upper
> sideband. You then do a WCSALIGN to align A and B. Should WCSALIGN take
> account of the differnce in sideband or not? That is, should WCSALIGN
> remap B so that the axis represents the lower sideband?
WCSALIGN should not align sidebands, it should align the frequencies that
those sidebands refer to. I think this simplifies things since it means
that the DSBSpecFrame subclass does not have to be coupled with a
DSBSpecMapping subclass. In a scheme where a DSBSpecFrame adds three new
attributes (the IF, the observed sideband and the current sideband) I
think that changing the current sideband should just result in different
frequencies being returned - all other transformations from frequency
remain the same. I don't know enough about AST to comment on
the feasibility of this approach.
To be more concrete, if I have a DSB spectrum using the lower sideband and
I want to generate a long spectrum consisting of both sidebands (with a
gap in the middle of course) I would copy the first spectrum, change the
current sideband to USB and then run WCSALIGN. If sidebands were aligning
I would end up with two identical spectra again, if sidebands were not
aligning I would end up with two spectra but one would be offset and
flipped by approximately 2*IF.
I'm not sure whether the aligned spectrum should still be a DSBSpecFrame
or whether it should now be just a SpecFrame.
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|