[daip] CVEL & SETJY questions (Doppler tracking correction)

[Eric Greisen]

Michael Bietenholz wrote:
> I was helping a student (Alet Dewitt) with AIPS and we were having
> some difficulty with using CVEL to apply the manual Doppler tracking.
> 
> Your instructions say to use SETJY to set the rest-freq. and then
> run CVEL.
> 
> We did this, entering the frequency in RESTFREQ.  However, when we
> then ran CVEL it caused large shifts (>100 channels; outside of our
> observed band)
> 
> Trying to track down what was going on , I printed out the SU table
> with PRTAB.  Now what is not quite clear to me is what the entries in
> the SU table that are made by SETJY are, in particular, what is the
> "LSRVEL" is.  The things I would have thought it might be is either
> the velocity of the telescopes (ie. array centre) wrt. the local
> standard of rest, or effective velocity offset implied by the
> difference between the observed centre frequency (once corrected for
> earth's motion wrt. the LSR) and the rest-frequency of the line.
> 
> As an experiment, I set RESTFREQ to exactly the observed centre
> frequencies in SETJY (ie. Header freq + FQ table offset per IF).
> SETJY then fills in two *different* values for LSRVEL for the two
> IF's, which doesn't seem right in either the above interpretations of
> LSRVEL.
> 
> So, what we've got I think is observations for a maser, made
> with Doppler tracking turned off (because of EVLA).  We just want
> to correct for the small time-dependent frequency offsets
> caused by the motion of the array.  What do we enter for
> RESTFREQ, SYSVEL, VELTYP and VELDEF in SETJY?
> 
> Also - we have both continuum calibrator sources and a maser
> source - does one apply CVEL to both or just the maser?
> (seems like it shouldn't make much diff to the calibrators
> which will have smooth spectra)

I am sorry to be slow in responding to this.  The critical parameter in 
SETJY is clearly the velocity and frequency axis pixel of that velocity.
Since multiple IFs are at different velocities (assuming the same 
spectral line is relevant in each case), you will need to run SETJY 
separately for each IF.  CVEL is one of the few tasks in AIPS that is 
smart enough to look at CL table one (assuming you have that file as 
produced by FILLM - not by INDXR or by some non-FILLM path such as 
through CASA) to get the full frequency information including any 
changes made over time to adjust the data to approximately (or exactly) 
a tracked velocity.

The code in CVLEL is simple enough although this hides a bunch of 
complex details that got the current obs frequency and current
dopler offset of the telescope.
C                                       Total velocity of
C                                       reference pixel
              VELTOT = VEL(IFNO) + DOPVEL
C                                       Frequency of reference pixel
C                                       in desired frame of reference
              IF (.NOT. RADIO) THEN
                 FREQTO = RSTFRQ(IFNO) / (1.D0 + VELTOT / VELITE)
              ELSE
                 FREQTO = RSTFRQ(IFNO) * (1.D0 - VELTOT / VELITE)
                 END IF
C                                       Required pixel shift
              DELI = (OBSFRQ(IFNO) - FREQTO) / FINT

Large shifts imply the the velocity parameters you have given it - what 
velocity you want at a specific spectral channel - were rather different 
than the actual velocity of that channel to start with.

Eric Greisen