[fitsbits] frequency definitions

[Maren Purves]

Tom Kuiper wrote:
> Mike Nolan wrote:
>> FREQREST should be the line frequency, but some writers put it in as the
>> center of the band to make some piece of software happy: Some readers need it,
>> but if you're doing a raw search, it doesn't make a lot of sense.
>>   
> We had a situation at CSO  recently in which we wanted two specific 
> lines (one in each sideband) well positioned in the spectrometer.  We 
> calculated a frequency that would do that.  That was the frequency which 
> was Doppler-shifted to tune the receiver.  I would call that FREQREST, 
> i.e., the frequency we wanted at the center of the band in the source 
> rest frame.  However, I agree that this makes a velocity scale hard to 
> interpret.

but ...

your 2 sidebands will have different frequencies.

And my interpretation is that the _rest_ frequency is in lab context,
or in the rest frame of the _source_.

It doesn't matter if you put in an extra keyword or 3 to satisfy
the requirements of one or the other data reduction package,
redundant as they may be, as long as they are consistent.

To chime in on the question asked about the radial velocities:
that's why you should specify the frame. A radial velocity
without specified frame doesn't have much meaning unless it's
rather large.


Aloha,
Maren Purves
(these days at UKIRT, but done my share of radio and submm)