Affiliation
Variable Stars South (VSS)
Sat, 07/13/2019 - 04:33
Hi
I have had problems with a few images trying to measure a star with a close companion.
Would it be possible to add a feature where the target can be moved slightly to better centre the star.
See the image attached for the problem. Cleary I can not use the measurement made for this particular star.
Thanks
Terry
File Upload
VPHOT.PNG121.7 KB
Hi Terry,
In the case of a close, nearly equal brightness, double like you've shown, the best aperture photometry approach is to use the Nemesis spreadsheet by Lew Cook:
https://www.aavso.org/software-directory
(near the bottom). It does require knowledge of the companion star brightness, of course. I think the original VPHOT did a "snap to" to the brightest star within a given radius, but that is usually the wrong approach (most target stars seem to be the faint star of the pair!). You may be able to take the sequence that you create and save, and modify the coordinates; I know that you can change the magnitudes so maybe this next level of editing is available.
Close doubles are difficult. The best measuring tools are either psf-fitting photometry like DAOPHOT or PSFextractor, or the complete image subtraction method pioneered by Alard. My recommendation is to avoid close doubles and choose another, equally interesting, object.
Arne
Thanks
I will try both methods described.
Unfortunately Arne sometimes close doubles are unavoidable. In this case it is a nova in the SMC and I need the photometry to flux calibrate spectra.
Terry
Terry:
There are a few things to try. Note that VPhot has a pixel/adu mode (above image). When you select it and click on the target of interest, you will zoom in on the two companion stars. You can move the aperture slightly to place the aperture on the correct target and only that star. Of course, the aperture needs to be small enough to exclude the companion. Give this a try. If the two stars overlap too closely as a result of the inadequate resolution achieved with your scope, you may need to give up BUT, give it a try and visually confirm the separation.
You can also modify (reduce) the aperture and sky annulus settings AND the centroid determination under Tools and Tools/Settings. Be sure to confirm that the settings look appropriate on the image!
HTH, Ken
Hi Terry,
If you want to VPHOT-share the images to HQA, preferably multifilter (but I can deal with a single bandpass), I'll go ahead and psf-fit them so that you have a magnitude for the companion - and then you can use Nemesis.
Arne
Terry:
This sounds like an interesting experiment. Could you also share images to me (MZK)?
It will be interesting to compare magnitudes obtained by both PSF and aperture photometry?
Ken
I have shared the images.
I have also measured the star using AIP4WIN at home. It will be interesting to see the difference.
Terry
Terry:
I ran the V image. The stars are actually almost completely separated, but not quite. I reduced the aperture to a radius of 2 pix. Smaller than I normally would use but not completely unreasonable. I increased the gap to include the companion completely and keep it out of the sky annulus. I used all aavso comps (>SNR=20) except the faintest. The V magnitude is that reported for the right most image in the attached screen shot. I then moved the aperture in pixel adu mode just a little by adjusting the position numbers. That magnitude is the middle image. I then used the pix/adu mode to move the aperture with the adjustment arrows. That moved the aperture visually a bit to far to the lower left and includes too much background. That magnitude is from the left most image and is clearly fainter than it should be?
In any case, the magnitudes only differ by a max of 0.06 mag across all three attempts. I'd say you are close enough for your need?? What did you get with AIP4WIN?
Ken
Sorry I'm slow replying bu my internet has been down.
AIP4WIN measured V=12.72 using an ensemble with an error of 0.01.
When I moved the annulus in Vphot I measured 12.835using a 2 pixel annulus so a significant diference between the 2 methods.
If I use a 3 pixel annulus it becomes 12.756 so much closer. Fortunately for my spectra flux calibration it probably makes little difference as long as I use the same technique for all of them.
Thanks
Terry
Hi Terry:
I think the most reassuring thing is that you got the same result as I did with aperture radius=2 in VPhot.
Do you know what aperture was used in AIP4Win? Perhaps about 3? Aperture=2 will include less impact from the companion and is probably closer to the truth?!
Hope you learned a little about the adu/pixel mode in VPhot. I think it helps for this close companion.
Maybe Arne will get a chance to run the images with PSF-fit?
Ken
Sorry I was so slow in getting these results. I had to make some changes in IRAF, as I have not done any psf-fitting since we changed computers at HQ. These results are psf fit, using four nearby comp stars (122, 127, 133, 138).
Star V Rc Ic
Target: 12.847 0.027 12.112 0.016 11.809 0.017
StarSE: 12.764 0.026 12.254 0.018 11.777 0.015
separation in X: 3.33 pixels
separation in Y: 2.96 pixels
Note that fwhm is about 2 pixels in these images, so the wings do overlap, which makes aperture photometry quite difficult. The brighter Rc value is probably due to Halpha emission.
Arne
So agreement was not too bad! Within error.
VPhot (TCB168) = 12.835 ??
VPhot (MZK) = 12.829 0.015
PSF (HQA) = 12.847 0.027
This was a pretty easy case for psf fitting, as both stars are about the same brightness and they are nearly split. The difficult ones are when the stars are closer together, and/or when one star is significantly fainter than the other. Then a 0.02mag error on the bright star means a 0.2mag error on the fainter one, for example. That said, I've gotten surprisingly good light curves for cases completely undoable with aperture photometry - but with a lot more effort.
This nova will become MUCH more difficult to measure properly, as it fades closer to quiescence. The SMC is a very crowded field, and contaminating background stars will start to flatten out the fade pretty soon.
Arne