Accurate bandpasss/flux calibration at 7mm

Got a calibration problem? Discuss it here.

Moderator: Mark.Wieringa

Accurate bandpasss/flux calibration at 7mm

Postby tg1989 » Thu Oct 20, 2016 11:38 am

Hi all,

I have (yet another!!) calibration recipe question. You may be sick of them just as much as I am! Sorry for the wall of text. If I have been unclear please let me know and I will attempt to rephrase.

The context is much the same as my earlier queries. I have a wide distribution of program sources, each with their own secondary calibrator. As we are observing at 7mm, we are to use Uranus to provide the flux density scale. 1921 was used to provide a bandpass calibration. A H-array was used during the observing, and to cut down on the overheads we observed each program source for a straight 10 minutes with the intention of fitting directly to the visibilities bypassing the imaging process. A consequence of this is that we have sparse uv coverage and using options=qusolve in gpcal will produce a degenerative solution.

As we are using a planet, mfboot is to be used to perform the actual flux scaling correction.

The main question I have. Since we are using some source that isn't 1934 to provide a bandpass calibration solution, mfcal assumes a flat spectrum across the band. In order to construct a more accurate flux calibration scale (as outlined in the ATCA user guide) the suggested approach at the 15mm band (where 1934 can be used as a flux calibrator) is to:

    mfcal vis=1921-293 interval=0.1
    gpcal vis=1921-293 interval=0.1 nfbin=2
    gpcopy vis=1921-293 out=1934-638
    gpcal vis=1934-638 interval=0.1 nfbin=2
    gpboot vis=1921-293 cal=1934-638

At this point we can use uvfmeas and plot the two CABB IFs (whatever they may be) and fit a flux model that may be used by mfcal to solve for the bandpass while accounting for the spectral slope of a source.

    uvfmeas vis=1921-293.IF1,1921-293.IF2 order=1 options=log,mfflux
    mfcal vis=1921-293 flux=<string_from_uvfmeas> interval=0.1
    gpcal vis=1921-293 interval=0.1 nfbin=2

Here mfcal/gpcal would be performed on both of the IFs separately. Once completed then 1921-293 is essentially a flux calibrator and typical calibration is then carried out with the appropriate phase calibrators.

    gpcopy vis=1921-293 out=<phase>
    gpcal vis=<phase> interval=0.1 nfbin=xyvary,qusolve
    gpboot vis=<phase> cal=1921-293
    gpcopy vis=<phase> out=<source>

So - I am wondering what is the best way to translate the above procedure to use Uranus as the flux calibrator? My current approach is more or less identical, with the exception that:

    gpboot vis=1921-293 cal=1934-638

is replaced with:

    mfboot vis=1921-293,uranus select=source(uranus)

In my head, once 1921-293 has the flux calibration of Uranus applied, then it is a 'normal' flux calibrator and gpboot may be used subsequently when calibrating the phase calibrators, as the appropriate nfbin options have been used in gpcal. I understand that the reason why we use mfboot is that it implements Uranus' model, where as mfcal/gpcal doesn't. Since it applies the correction to the calibration tables (and not the visibilities per its help page) gpboot/gpcopy should be OK. is this reasonable? Or should the task:

    gpboot vis=<phase> cal=1921-293

be replaced with:

    mfboot vis=<phase>,1921-293 select=source(1921-293)

A secondary question is is it 'safe' to use exclusively options=xyvary throughout the calibration recipe, as introducing qusolve only gives a failed convergence. I am not really interested in the polarization properties of these sources.

If there are any other comments/suggestions please let me know. The only other alternative I could think of was to not split the IFs. Treat the dataset for each source as a single UV file. I was unable to get to far with this as I did some frequency switching throughout the run. Although I uvcat'ed the appropriate 7mm IFs together after atlod/uvsplit, the visibilities were not in time order and gpcal fell over.

As any way where I can plot the model of Uranus used by Miriad?

Posts: 17
Joined: Sun Dec 07, 2014 3:04 pm

Re: Accurate bandpasss/flux calibration at 7mm

Postby ste616 » Thu Oct 20, 2016 1:30 pm


We do recommend now that 1934-638 be used at 7mm, as its flux density scale is much better than Uranus in that band now. But since you probably didn't make an observation of 1934-638 during your observations, we'll stick with Uranus.

We'll answer the easy secondary question first: omitting qusolve, and thus not attempting to solve for the leakages will produce a small (percent level) error in Stokes I measurements. Since the accuracy available at 7mm does not reach that level anyway, polarisation calibration is not critical.

You are correct to replace gpboot with mfboot when bootstrapping 1921-293 with Uranus. However, once you've fed the flux density model back into mfcal for 1921-293, you should consider it to be a point source flux density calibrator, and thus using gpboot for the second round of calibration is the preferred method. The mfboot task can of course be used, but it is not necessary, and gpboot is probably safer.

To plot the flux density of Uranus for a particular date, frequency and uv range, use the task plplt.
Jamie Stevens
ATCA Senior System Scientist
Site Admin
Posts: 219
Joined: Thu Feb 04, 2010 3:27 pm
Location: Paul Wild Observatory Narrabri NSW

Return to Calibration

Who is online

Users browsing this forum: No registered users and 1 guest