7mm Bandpass Spectrum - IFs not aligning

Got a calibration problem? Discuss it here.

Moderator: Mark.Wieringa

Post Reply
tg1989
Posts: 17
Joined: Sun Dec 07, 2014 3:04 pm

7mm Bandpass Spectrum - IFs not aligning

Post by tg1989 »

I am reducing some recent data at 33 and 35GHz I observed in a H168 array.

As 1934-638 is not bright enough at these frequencies, I used a different source to perform the bandpass calibration step with. On this particular day this source was 0420-014.

As 0420-014 does not have a spectral model built into miriad, mfcal assumes a flat spectral shape when solving for the bandpass solution - see the '0420-014_mfcalgpcal.png' file. Following the process outlined on the ATCA User Guide, the true intrinsic shape of 0420-014 can be estimated and used as part of the bandpass calibration by:
(1) - Copying the initial bandpass/gain solutions across both IFs from 0420-014 to 1934-638
(2) - gpcal 1934-638 across both IFs
(3) - gpboot 0420-014 with 1934-638
(4) - use uvfmeas on 0420-014 to constrain a spectral index compatible with mfcal flux keyword

In the past this has worked well for me, where after step (3) I get the two IFs following a consistent slope. With these data though I am getting something that I am having trouble making much sense of. If you refer to the '0420_mismatch.png' file I attach, you can see that the two IFs are not exactly being consistent with one another.

An oversight when observing, I had the pointing mode set to 'point' and not 'dwell' in the schedule file for both 0420-014 and 1934-638. Hence, when tracking these sources all data were being flagged. I can get around this in miriad by using atlod with the caldata keyword, and subsequently manually flagging the pointing update data. I don't think is related to what I am seeing in the attached plot, but thought I'd share just to be sure.

I am wondering if anyone has any suggestions on what to do? Is this is as good as it is going to get, or is there something else that I should be doing?
Attachments
0420_mfcalgpcal.png
0420_mfcalgpcal.png (9.83 KiB) Viewed 6479 times
0420_mismatch.png
0420_mismatch.png (10.09 KiB) Viewed 6479 times
ste616
Site Admin
Posts: 220
Joined: Thu Feb 04, 2010 3:27 pm
Location: Paul Wild Observatory Narrabri NSW

Re: 7mm Bandpass Spectrum - IFs not aligning

Post by ste616 »

Hi Tim,

Although that method is pretty good, it can only calibrate for certain effects. If, for example, you don't do opacity correction during the atlod stage, I'm not sure this calibration method will compensate.

Now, even if tell atlod to do the opacity correction, it may not apply it to the flagged data; that's my concern. Is there another source that you could try this with, in your dataset, that didn't have to go through the unflagging process?
cheers
Jamie Stevens
ATCA Senior System Scientist
tg1989
Posts: 17
Joined: Sun Dec 07, 2014 3:04 pm

Re: 7mm Bandpass Spectrum - IFs not aligning

Post by tg1989 »

Hi Jamie,
Although that method is pretty good, it can only calibrate for certain effects. If, for example, you don't do opacity correction during the atlod stage, I'm not sure this calibration method will compensate.
So in atlod I do use the opcorr option with nopcorr set to 16.
Now, even if tell atlod to do the opacity correction, it may not apply it to the flagged data; that's my concern.
I thought the easy answer to this would be to read the atlod.for code, which I did. I also typed up some notes. And then I hit the back button accidentally and lost it... But as best I can tell...

Opacity correction appears to be done in the subroutine PokeFlsh on lines 1702 through 1747. Likewise, the flagging of calibration data is performed in RPDisp on line 2734. I think these work on an integration basis. If PokeFlsh (or AtFlush, which calls PokeFlsh) are called, then that integration has its opacity corrected and written out to the output dataset. I think this is done in the RPDisp subroutine on line 2937 after a last check.

So, I think if the data written out is written out to a file, then it is opacity corrected if the opcorr option is activated. But I am no solid Fortran developer, and reading Miriad source is hard!
Is there another source that you could try this with, in your dataset, that didn't have to go through the unflagging process?
For that day no there is not.

Looking at the second day of observing, where this observing mode mistake in the schedule file was fixed, I am seeing similar behavior. The attached uvfmeas plots show the spectra of a different bandpass calibrator as I am performing the calibration steps. You can see in the 1034293_postgpboot.png file that the two IFs are pretty different.

Increasing the nfbins from 2 to 4 for this day for the 33GHz band I am getting a different shape - one that could be believable? See the 1034293-nfbin4-33ghz.png file. Is it worth trying the whole process with a larger collection of nfbin's? I know in the past you have had good success with only needing two.

As an aside, I am curious what the 'bumpy' feature is at ~33.8GHz.
Attachments
Bandpass calibrator after gpboot'ing with 1934-638
Bandpass calibrator after gpboot'ing with 1934-638
1034293_postgpboot.png (8.63 KiB) Viewed 6474 times
Bandpass calibration after mfcal/gpcal
Bandpass calibration after mfcal/gpcal
1034293_pregpboot.png (9.49 KiB) Viewed 6474 times
Raw data without calibration
Raw data without calibration
1034293_raw.png (10.21 KiB) Viewed 6474 times
tg1989
Posts: 17
Joined: Sun Dec 07, 2014 3:04 pm

Re: 7mm Bandpass Spectrum - IFs not aligning

Post by tg1989 »

And the image I referenced that I could not attach to the previous post.
Attachments
1034293_nfbin4-33ghz.png
1034293_nfbin4-33ghz.png (8.74 KiB) Viewed 6474 times
Post Reply