[daip] Frequency/velocity axis for FITS image cubes produced from VLA data

[Eric Greisen]

Jesus Martin-Pintado wrote:
> Dear Dr. Greisen,
> I am writing you because I am running into the problem of understanding 
> the frequency/velocity axis for FITS image cubes produced from VLA data 
> using AIPS. I know from your publications that you are one of the 
> world's expert on FITS and you also know AIPS very well. I hope you can 
> help me on this issue.
> 
> I would like to reconstruct the frequency/velocity scale to be able to 
> derive the observed parameters of different lines across the frequency 
> band. I have a problem when I read the VLA fits file in FREQ or VELO.
> 
The topocentric (actual observed) frequency is required for imaging and 
the like so that is what we keep.  It scales the baselines into 
wavelengths.  You can compute any other reference frame frequency by 
computing the velocities at the time wrt that frame.  In the example in 
Paper III, I did that by hand using non-AIPS software or a calculator 
actually.

> In a cube with the FREQ axis, it seems that one gets the topocentric 
> (sky) frequency.  I though that one could get the frequency axis 
> relative to the source velocity frame as defined by VELREF by using the 
> RESTFREQ, the ALTRPIX and the ALTRVAL as follows:
> freq(n)=RESTFREQ+(n-CRPIX3)*CDELT3
> velo(n)=ALTRVAL-(freq(n)-RESTFREQ)/RESTFREQ (for ALTRPIX=CRPIX3)
> 
If it were only so simple..  The velocity increment at the alternate ref 
pixel is

DVel =   - DelF (c + RefV) / F(altrefchan)  in optical convention
DVel =   - DelF (c - RefV) / F(altrefchan)  in radio convention

then in optical convention

Vel = RefV + DVel * (n - Nref) / (a + A * (n - Nref))
where A = -DevV / (c + RefV)
and   c = speed of light

The restfreq actually is never used.

> However, when I use the same cube but extracted with the VELO axis, I 
> get that the CDELT3 is not consistent neither with converting the 
> channel separation in frequency obtained from the same cube with the 
> FREQ axis using the frequencies from ALTRVAL or RESTFREQ.
> 
> Is there a way to get the information I need from the FITS header?.
> 

Possm will list the frequencies in a text file output.

Eric Greisen