Really three questions/issues: 1. Suppose I have a variable with a close companion, such as W Her. To minimize the flux contribution from the companion I set the aperture to 3 pixels and also shift the centroid one pixel right and one up (away from the companion). Then save the sequence. So far so good. Next time I bring up a W Her image and select my sequence, I notice the 3 pixel aperture is still set but the centroid shift is not. So, the first question is "is this expected?" or should the centroid shift have been saved along with the aperture setting? 2. This brings me to my second question. Knowing that I should adjust the centroid to get a bit away from the close companion, I don't see how to do this when transforming as the process uses the saved sequence which, as I mentioned before, apparently doesn't save the centroid offset. If I view the images in the report and change the centroid it has no effect on the results. So I need a way forward in shifting a centroid when doing a two-color transform (someone must have encountered this situation by now) 3. Finally, when doing a two-color transform in the case of a close companion, is it better to select a fixed aperture of 3 pixels (in this case) or use the automatic aperture?
Hi Scott:
I suspect you may recall from the VPhot Course that targets with very close companions are difficult to analyze. Expecting the two color transform tool or the time series tool to work well without individual manipulation is probably asking too much. Of course, it does depend on how close the companion is.
1. Since VPhot measures the centroid of the target (and comps) on each image, saving a different and not well defined aperture position would be really difficult to code in VPhot. You should not call this new position a centroid. It is clearly consistent with some other adu count than the true centroid adu maximum. Since your revised position does not match the actual stellar profile center on the different image, it is not the new centroid. What would the shifted aperture center correspond to compared to the true centroid? I'm sure you can understand that this would be difficult to reproduce consistently. Alternatively, you could shrink the aperture. Within reason, your aperture could include less of the star profile and still yield good magnitudes.
2/3. I think your other two questions are somewhat related. Other than recommending the option of a smaller aperture to exclude the companion, I recommend that you use the new TA function in VPhot. Analyze the two images (two filters) separately by carefully shrinking/shifting the aperture and running/saving the single image photometry report for each. Subsequently, open the TA link and choose the two individual Single Image Photometry reports in the top box. Click "Create Transformed AAVSO Report". You can then output/save the transformed report from the second box.
HTH, Ken
Ken,
Thanks for the reply. Yes, I recall your discussion in the VPhot course about handling close companions, but at the time I was just beginning to do transforms and hadn't thought about how to get transformed magnitudes in these situations. I was also unaware of how TA worked or in what situations it could/should be used, so your instruction on how to use it to get the transformed magnitudes is just what I needed. Having learned from you in the VPhot course the importance of being careful in making these measurements, I wanted to make sure I understand how to go about getting transformed magnitudes in these close companion circumstances to avoid reporting a magnitude excessively influenced by the companion.
P.S. Found this statement about the Transform Applier tool: "Please note: Ensemble observations are not transformable with this tool."
That would explain why I couldn't get it to work....I fed it ensemble measurements.
Regards,
Scott
Hi,
I have had similar problems and I have followed Ken's advice to get around it.
May I just show you what I have done for a sanity check?
The target is Z Tau which has a close companion. Moving the centroid is necessary. The centroid position is not saved in a sequence, so once it is set, one must generate a "Single Image Photometry" before loading another image. Using VPhot's "Two Color Transform" will not preserve the centroid position.
There are 2 calibrated and plate solved images, one in V and one in I.
For each of the 2 images the following was done:
1) A sequence was created with a check star and a single comp star chosen based on low err for the filter (not much more than .02) using
the Variable Star Plotter Field Photometry Database: (Catalogs->Show AAVSO Sequence). More than 1 comp star will create and "Ensemble" and this will not
work in VPhot's "Transform Applier" (TA)
2) The centroid was adjusted using Pixel/ADU Mode
3) A Single Image Photometry report was created and a AAVSO Report downloaded. This is not transformed
Then the, AAVSOReport_Z-Tau_I_20220119.txt and AAVSOReport_Z-Tau_V_20220119.txt, which were generated in step 3, are uploaded to the VPhot TransformApplier. This will create AAVSOReport_20220120_TA.txt which shows the transformed magnitudes and can then be uploaded as an observation.
Is this procedure reasonable?
Getting the centroid just right is problematic.
Thanks!
I think it's reasonable because that's what I do(!) As you say, I generate single image photometry report files where I have used the centroid-moving functionality to make an adjustment when that is necessary. I actually combine the two observation lines into a single report file first, and then use the Transform Applier tool to do the transformation and generate a new transformed report.
That's a lot more trouble than using the handy "Two Color Transform" tool, but it's the only way I know to deal with centroid issues where there are companions that throw the built-in routines off.
As an aside, thanks for working Z Tau and taking care not to just do photometry of the ~13.5 magnitude companion when Z Tau is faint! I worked with Arne and SRO to follow it down to minimum and published that in the AAVSO journal several years ago. I had to use IRAF and psf fitting. That was way more work that I had hoped it would be.
Clearest skies,
Walt
Walt,
I have been looking into astropy, photutils, and photutils.psf (running in an anaconda/python notebook) which has the DAOPHOT algorithm. (https://ui.adsabs.harvard.edu/abs/1987PASP...99..191S/abstract) I'm trying to build a framework using these so that one can measure stars with close companions. (I'm sure somebody in a grad school has done this already.)
When you used IRAF, was it more readily available at the time. Did you consider DAOPHOT? I'm finding that running IRAF on windows is problematical. (I don't want to give up disk space for a linux partition!)
Thanks!
Pierre "Pete" Fleurant
Hi Pete,
I started out with the standalone DAOPHOT (you could get the Fortran source on a magtape from Peter Stetson...), and then they incorporated DAOPHOT into IRAF, so I used it from there in later years.
I'm using IRAF from my Macbook pro, but using VirtualBox to create a separate virtual linux system on an SSD. I think VirtualBox is available under Windows, and that might be a solution. I had not seen photutils.psf; that might be an interesting approach. There is a Facebook group called Python Users in Astronomy, where you might inquire to see if a grad student has used that library for working with close companions. If you find an answer to that, let me know too!
Arne
Pete, Arne,
A friend has made me aware of another option for doing psf photometry. There is an astronomy education organization in Australia called Our Solar Siblings. They have put together what looks like a very powerful system that uses DAOPhot and five other software packages to reduce your images and send you the data as a service. I don't think I can include a link on this forum. Google Our Solar Siblings Pipeline to find it.
I spend huge amounts of time reducing data and am eternally behind. If this is as robust and capable as it seems, I will likely sign on.
-Walt
Hi
After catching up on what exactly PSF photometry is I started running a program I found.
There is a M Usatov who has written MetroPSF (latest version is 0.15). It is entirely in Python and uses python libraries: numpy, pandas, scipy, Astropy, pillow, astroquery, matplotlib, photutils,etc. It is open source.
Photutils contains a variant of DAOPHOT.
Check out: https://www.aavso.org/my-differential-ensemble-psf-photometry-program
I installed it on my Windows 10 machine and gave it a go on Z Tau.
I can’t post any screenshots on this thread but at least I can show some tabular results.
auid differential_mag mag_error auid_mag
000-BBK-305 14.11 0.008 14.077
000-BBK-305 17.025 2.923 14.077
000-BBK-310 12.619 0.067 12.673
000-BBK-310 15.987 3.301 12.673
000-BBK-311 14.999 0.037 14.934
000-BBK-311 17.96 2.998 14.934
000-BBK-315 13.587 0.012 13.585
000-BBK-315 16.953 3.378 13.585
000-BBK-315 15.027 1.452 13.585
000-BBK-316 16.187 0.105 16.075
000-BBK-316 19.383 3.301 16.075
000-BBK-320 11.849 0.033 11.81
000-BBK-321 12.985 0.031 12.992
000-BBK-321 16.015 2.999 12.992
000-BBK-323 16.441 0.007 16.31
000-BBK-323 19.468 3.02 16.31
000-BBK-324 16.833 16.726
000-BBK-324 19.685 16.726 <------? dont know exactly why these repeat (detected on diff iterations?)
000-BBK-327 14.467 14.434
000-BBK-327 17.482 14.434
000-BBK-329 15.944 15.854
000-BBK-329 19.066 15.854 <------?
000-BBK-329 15.854
000-BBK-330 14.644 0.074 14.546
000-BBK-330 17.919 3.349 14.546
000-BBK-335 16.708 16.576
000-BBK-335 19.804 16.576
000-BBK-337 15.521 0.138 15.385
000-BBK-337 18.701 3.318 15.385
Unlike VPhot, with MetroPSF I can’t easily remove the outliers.
Z Tau, 2022-01-19 03:09:30.470 UTC, V = 17.29 ± nan
ra_fit dec_fit vsx_id differential_mag
88.10378797 15.79569901 Z Tau 17.285
I am learning how to remove the outliers seen in the linear regression model, and get some number for err, (and not ‘not a number’).
I must say though, it was cool to see as I increased the number of iterations of the PSF Photometry, Z Tau was "discovered"!
Thanks!
Pierre “Pete” Fleurant
Hi Walt,
This sounds good, but is a commercial service at between $15 and $55 per month. You can use the AAVSOnet pipeline for free, if the images were taken with AAVSOnet telescopes. We have considered making the pipeline available as freeware some day, but have some features to add before that happens.
Let us know how the system works, if you go ahead and sign up!
Arne
Thanks Arne. I'd love to be able to use the AAVSOnet pipeline for my own observatory down the road. These folks' system might be the answer in the meantime. It costs money but it would be worth it if I could dig out from under some of my backlog and reduce reduction time for new data. I'd be glad to pay for use of the AAVSOnet pipeline in the future when it's ready.
-Walt
Pete,
Yes, I used DAOPhot within IRAF on a linux box I had put together at the time. It was quite the adventure back then to get IRAF installed. Sadly, that linux box has since died. Looks to be a motherboard problem.
-Walt