[Aavso-photometry] Effect of aperture size

[arne]

Jim Roe wrote:
> 
> arne wrote:
> 
> 
>>A little more detail?  Are you saying that the differential magnitudes
>>like (V-C) were offset by 0.3mag between the 6pixel radius and the
>>9 pixel radius aperture sizes?  This should not happen.  If you mean
>>the instrumental magnitude like Vins or Cins, then those will change
>>with aperture size since the star profile extends quite a distance from
>>the peak pixel; usually larger apertures will give a brighter magnitude
>>as you include more of the starlight.  If you mean that the (V-C) 
>>differential
>>magnitude differed by 0.3mag between apertures, then generally that can
>>only happen if you include companion stars within the aperture, or
>>if you are changing other things like the sky annulus simultaneously.
> 
> 
> I went over the data and (V-C) in both cases was essentially the same. 
> The only difference was that all of the instrumental magnitudes shifted 
> by the same amount (0.273 mag fainter).  The only thing I could imagine 
> that would cause this would be a different background level subtracting.
> 
> As I merely increased the aperture radius, the guard width stayed the 
> same so the outer annulus increased in radius AND area.  Going back to 
> the images and decreasing the guard band width to put the outer annulus 
> at the same location eliminated the discrepancy.  HOORAY!  But what does 
> that say about the uniformity of the background over pixel changes?  It 
> seems to be uniform - I did the flat field corrections.  My images 
> drifted a bit over the 2+ hour period and, at one point, I re-centered. 
>   If there were a flat field problem, one would think it would show up 
> more severely than a consistency to 3 mmag in the results?
> 
> AND that doesn't solve my essential problem in that how do I choose 
> where to put the background sampling annulus for the rest of my 
> measurements now and in the future (avoiding companion stars is 
> obvious)? :-(  If I try several settings and get different results, how 
> could I choose?
> 
(V-C) is by far the important quantity: it should remain the same, no
matter what your aperture size, as you are measuring the same fraction
of starlight for both the variable and its comparison star.  Which aperture
you choose depends on the signal/noise and what contaminating stars
might be contained within the aperture.  See Howell 1990ASPC....8...312H
for more detail.

The sky annulus is usually pretty robust against contaminating stars,
as rejection algorithms are used to clean its pixels.  Its size should
not significantly change the sky level unless the sky has a strong
gradient or structure.

What it looks like, and this may be a use-of-software problem, is that
the sky is not being determined correctly.  I'd test by keeping the
same measuring aperture, and then growing the sky annulus and see if the
measured magnitude changes significantly.  Do this on an image where
you are seeing the 0.273mag change.  For example, the program may not
be calculating the area contained within the sky annulus properly.
Then try the opposite experiment: use a large sky annulus, and
increase the measuring aperture.  For a well-exposed star, you
should see the curve-of-growth plot, where the magnitude increases
rapidly at first as you expand the aperture radius, and then slows
down to asymptotically approach a limit.

If you are still confused after this, send me a flatfielded fits
image and I'll examine it for you.
Arne