[Aavso-photometry] photometry variation with measurement aperture

[arne]

Walt Cooney wrote:

>Folks,
>
>I could use a little advice.   I am doing some standardized photometry on DK CVn
>and have found I can move the magnitudes up or down by a few hundredths
>depending on what measurement aperture I use.  Clearly that's bad for
>standardized photometry.  Aperture size appears to be an optimization between
>S/N for seeing the trend behavior vs. producing data with good absolute
>accuracy.
>
>I'd like to understand why it matters.  If you carve off the top of the light
>curve distribution function, I don't see why is should matter whether you carve
>off and measure more or less as long as you carve off and measure the same for
>each star.  It can't (?) be a problem with few pixel statistics because that
>would show up as random noise.
>
>I'd appreciate some thoughts on this.
>
>  
>
Walt, give us more details.  Software program?  fwhm in pixels of the 
stars?  Size of
the apertures, and magnitude change?  Inner/outer sky annuli radii?  
Counts in the
aperture and sky?
  In general, you will get different results as you change the aperture 
size, just based
on random statistics.  As the aperture gets smaller, you run into 
"rounding" problems,
where there are differences due to including partial pixels in an 
attempt to create
a circular aperture.  In addition, for faint stars, as the aperture gets 
smaller, the
signal/noise generally increases as you are including fewer sky pixels 
in the aperture,
so the measure tends to asymptotically approach the correct answer.
Arne