[fitsbits] frequency definitions
Tom Kuiper
definitions are. Is the following right?
If there is a place where all this is clearly laid ou, please refer me
to that.
CRVALx - the value corresponding to the pixel CRPIXx in frame
defined by CTYPEx.
OBSFREQ - the actual frequency to which the receiver is tuned
corresponding to the center of the IF.
RESTFREQ - the receiver frequency (corresponding to the center of the
IF) in the reference frame defined in TTYPEx.
( Vdop )
OBSFREQ = RESTFREQ * ( 1 - ------)
( c )
where Vdop is the velocity of the reference frame with respect to the
observer.
If we assume that CRPIXx is the pixel corresponding to the center of the
IF, then
CRVALx = RESTFREQ if CTYPEx = FREQ-LSR,
CRVALx = OBSFREQ if CTYPEx = FREQ-OBS
What is the difference between FREQ-OBS and FREQREST?
Thanks and regards
Tom
LC's No-Spam Newsreading account
> I'd like to get it clear in my mind what the various frequency
> definitions are. Is the following right? If there is a place where
> all this is clearly laid ou, please refer me to that.
Not a spectral WCS expert, but the standard and official references
are :
- http://fits.gsfc.nasa.gov/fits_standard.html
the latest release of the FITS standard, in particular chapter 8
- the full WCS paper III, either as published or
http://www.atnf.csiro.au/people/mcalabre/WCS/scs.pdf
(the other WCS papers are all listed in
http://www.atnf.csiro.au/people/mcalabre/WCS/index.html
--
----------------------------------------------------------------------
nos...@mi.iasf.cnr.it is a newsreading account used by more persons to
avoid unwanted spam. Any mail returning to this address will be rejected.
Users can disclose their e-mail address in the article if they wish so.
David Berry
covered by paper III. OBSFREQ, etc, are not in the standard. There are
probably various different schemes in use out there. For instance JCMT
and CLASS software use a system based on a keyword called IMAGFREQ,
which holds the image sideband rest frequency (with RESTFREQ holding
the observed sideband rest frequency).
David
2009/5/5 LC's No-Spam Newsreading account <nos...@mi.iasf.cnr.it>:
> _______________________________________________
> fitsbits mailing list
> fits...@listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/fitsbits
>
Tom Kuiper
Regards
Tom
David Berry wrote:
It looks like Tom's query was about dual sideband data, which is not covered by paper III. OBSFREQ, etc, are not in the standard. There are probably various different schemes in use out there. For instance JCMT and CLASS software use a system based on a keyword called IMAGFREQ, which holds the image sideband rest frequency (with RESTFREQ holding the observed sideband rest frequency). David 2009/5/5 LC's No-Spam Newsreading account <nos...@mi.iasf.cnr.it>:
On Mon, 4 May 2009, Tom Kuiper wrote:
is a newsreading account used by more persons to avoid unwanted spam. Any mail returning to this address will be rejected. Users can disclose their e-mail address in the article if they wish so. _______________________________________________ fitsbits mailing list fits...@listmgr.cv.nrao.edu http://listmgr.cv.nrao.edu/mailman/listinfo/fitsbits
Bob Garwood
in here.
As mentioned elsewhere, the primary WCS papers are referenced in that
previous reply and cited below. The SDFITS note in aips++ predates all
of that. SDFITS has a long history, which I won't go into here. It
remains primarily a locally implemented standard with a few version of
"local" that aren't quite in agreement for lack of effort primarily on
my part to try and reach a real consensus and, most importantly,
exchange data between "local" implementations.
Prior to the WCS papers, lots of people obviously described a
frequency-like axis. So you're going to find that historically there
are keywords and columns for which you'll just have to talk to the folks
that wrote the FITS file to be sure you go the interpretation right - if
there is any ambiguity in what you find there as compared to the WCS papers.
For your specific questions ...
CRVALx - I'd say "yes"
OBSFREQ - 'probably'. I think "center of the IF" is possibly ambiguous
enough that you might want to double check with the author/writer
RESTFREQ - no. I'd interpret that as equivalent to the "RESTFRQ" value
described in the WCS papers. It's the rest frequency of the spectral
feature of interest - i.e. the frequency that that feature has when
observed at rest with respect to the thing producing that feature. It's
the lab-measured line frequency. It may or may not correspond to the
center of the IF, depending on how the IF was set up and if doppler
tracking was being done during the observations. It should only be
necessary/useful if you want to convert from a frequency to a velocity
(I'm 99% sure of that statement, but I'm sure someone here will correct
me if that isn't true) - or rather, this quantity isn't necessary to
recover the frequency at each pixel nor is it necessary to convert from
a frequency in one reference frame to a frequency in a different
reference frame assuming that TTYPEx is a frequency axis.
Then you ask" What is the difference between FREQ-OBS and FREQREST?".
But I'm not sure what you mean by "FREQ-OBS". You should an equation
earlier (badly formatted in my copy, but I think I understand what you
meant) showing the difference between OBSFREQ and RESTFREQ, so I don't
think you mean FREQ-OBS = OBSFREQ. SDFITS doesn't use FREQ-OBS so I
think you'll need to ask the creator of that FITS file what they
meant. Unless I missed something in a quick scan of the relevent WCS
paper, I don't see it defined there.
Cheers,
Bob
Tom Kuiper wrote:
> Well, not specifically about dual sideband data but about radio
> spectral line data in general. The best description of the relevant
> keywords that I've found is in the NRAO AIPS++ documentation at
> http://aips2.nrao.edu/docs/notes/236/node14.html. However, the
> definitions are not entirely unambiguous. That's why I posted my
> interpretation, to see if anyone disagreed.
>
> Regards
>
> Tom
>
> David Berry wrote:
>> It looks like Tom's query was about dual sideband data, which is not
>> covered by paper III. OBSFREQ, etc, are not in the standard. There are
>> probably various different schemes in use out there. For instance JCMT
>> and CLASS software use a system based on a keyword called IMAGFREQ,
>> which holds the image sideband rest frequency (with RESTFREQ holding
>> the observed sideband rest frequency).
>>
>> David
>>
>>
>> 2009/5/5 LC's No-Spam Newsreading account <nos...@mi.iasf.cnr.it>:
>>
>>> _______________________________________________
>>> fitsbits mailing list
>>> fits...@listmgr.cv.nrao.edu
>>> http://listmgr.cv.nrao.edu/mailman/listinfo/fitsbits
>>>
>>>
>>
>> _______________________________________________
>> fitsbits mailing list
>> fits...@listmgr.cv.nrao.edu
>> http://listmgr.cv.nrao.edu/mailman/listinfo/fitsbits
>>
>
> ------------------------------------------------------------------------
Tom Kuiper
> Then you ask" What is the difference between FREQ-OBS and FREQREST?".
> But I'm not sure what you mean by "FREQ-OBS". You should an equation
> earlier (badly formatted in my copy, but I think I understand what you
> meant) showing the difference between OBSFREQ and RESTFREQ, so I don't
> think you mean FREQ-OBS = OBSFREQ. SDFITS doesn't use FREQ-OBS so I
> think you'll need to ask the creator of that FITS file what they
> meant. Unless I missed something in a quick scan of the relevent WCS
> paper, I don't see it defined there.
>
value for TTYPEx. It refers to some kind of reference frame. I assume
it means this is a frequency axis defined in the frame of the observer.
OBSFREQ is a keyword and I confess that it's not clear to me why it is
needed if the same information, in some form, is required to define one
of the axes.
Best regards
Tom
Arnold Rots
First, there is no requirement that the reference pixel is at the
center of the IF band.
Second, there is no reason to assume that if TTYPE='FREQ-LSR' the
frequency at the center of the band is at the line's rest frequency.
More likely, it will be at the Doppler frequency of the object of
interest.
Third, if TTYPE='FREQ-OBS', it says that frequencies (OBSFREQ, if you
like) are expressed in truly observed frequencies;
if TTYPE='FREQ-LSR", the frequencies are labeled as if the observer
were moving with the LSR - i.e., the observer-to-LSR component has
been taken care of already, what remains is the objects Doppler velocity.
Fourth, although it looks like you are working on Galactic stuff in
the radio (and hence the radio definition seems appropriate), it may
be good to keep in mind that you formula:
( Vdop )
OBSFREQ = RESTFREQ * ( 1 - ------)
( c )
is only correct for Doppler velocities using the radio definition.
In the optical definition V = c*delta(lambda)/lambda0, rather than
V = c+delta(nu)/nu0.
And, of course, one may prefer barycentric over LSR velocities when
working on extragalactic data.
- Arnold
> Then you ask" What is the difference between FREQ-OBS and FREQREST?".
> But I'm not sure what you mean by "FREQ-OBS". You should an equation
> earlier (badly formatted in my copy, but I think I understand what you
> meant) showing the difference between OBSFREQ and RESTFREQ, so I don't
> think you mean FREQ-OBS = OBSFREQ. SDFITS doesn't use FREQ-OBS so I
> think you'll need to ask the creator of that FITS file what they
> meant. Unless I missed something in a quick scan of the relevent WCS
> paper, I don't see it defined there.
>
--------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701
60 Garden Street, MS 67 fax: +1 617 495 7356
Cambridge, MA 02138 ar...@head.cfa.harvard.edu
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------
Mike Nolan
> Bob Garwood wrote:
> > But I'm not sure what you mean by "FREQ-OBS". You should an equation
> > earlier (badly formatted in my copy, but I think I understand what you
> > meant) showing the difference between OBSFREQ and RESTFREQ, so I don't
> > think you mean FREQ-OBS = OBSFREQ. SDFITS doesn't use FREQ-OBS so I
> > think you'll need to ask the creator of that FITS file what they
> > meant. Unless I missed something in a quick scan of the relevent WCS
> > paper, I don't see it defined there.
> >
> value for TTYPEx. It refers to some kind of reference frame. I assume
> it means this is a frequency axis defined in the frame of the observer.
> OBSFREQ is a keyword and I confess that it's not clear to me why it is
As Bob suggested, it's an older word that likely means the same thing. FREQ-OBS
is defined in paper III (I think) to mean that the frequencies are as seen
at the telescope, uncorrected for (for example) reference frame. You can have
FREQ-other stuff as well in TTYPEx. SDFITS is quite a bit older and didn't have
all that flexibility, so it had single keywords like OBSFREQ.
Since FITS readers often want one thing or another, we often just put them
both in if it's needed.
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.
Cheers,
-Mike
--
Dr. Michael C. Nolan, Observatory Director
+1 787 878 2612x212 Fax: +1 787 878 1861 no...@naic.edu
Arecibo Observatory, HC 3 Box 53995, Arecibo, Puerto Rico 00612 USA
Tom Kuiper
> 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.
>
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.
Regards
Tom
Steve Allen
> And, of course, one may prefer barycentric over LSR velocities when
> working on extragalactic data.
This prompts me to ask.
Is there is a paper akin to
Lindegren and Dravins
http://adsabs.harvard.edu/abs/2003A%26A...401.1185L
which addresses the same sorts of issues of
"what do we mean when we say frequency/radial-velocity?"
In particular, a paper which addresses the issues with the contexts of
both the recent IAU definitions about precision of definitions and the
traditions of radio astronomy?
--
Steve Allen <s...@ucolick.org> WGS-84 (GPS)
UCO/Lick Observatory Natural Sciences II, Room 165 Lat +36.99855
University of California Voice: +1 831 459 3046 Lng -122.06015
Santa Cruz, CA 95064 http://www.ucolick.org/~sla/ Hgt +250 m
Bob Garwood
purpose. It's a useful thing to make note of - i.e. that's the
frequency we're using to do the updates necessary for dopper tracking,
but as you say, it would generate a not particularly useful velocity
scale and readers who are unfamiliar with the CSO and that use will be
likely to make that mistake, which I think is a bad thing.
And I'm sorry for mis-interpreting you use of "FREQ-OBS". In
retrospect, that's a CTYPEn value and Arnold's reply is useful. In the
SDFITS convention, OBSFREQ is explicitly defined to be in the observers
frame of reference independent of the CTYPEn value.
-Bob
Maren Purves
> 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)
Arnold Rots
The definition is mentioned in Eq. 8.
The only thing missing from the paper, as far as I can tell, is a
discussion of LSR - LSR-K and LSR-D.
- Arnold
> _______________________________________________
> fitsbits mailing list
> fits...@listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/fitsbits
>