[Aavso-photometry] Effect of aperture size

[Jim Roe]

I measured a transit of GJ436 in B.  The light curve was beautiful 
(IMHO), see

<http://brucegary.net/AXA/GJ436/gj436.htm#Light_Curves>

BUT, there was a discrepancy in the depth of the transit of some 3 mmag 
(compared to earlier results by lots of observers) so I re-measured the 
images using different aperture sizes (I use MaximDL). I have taken to 
using two or more reference stars (ensemble) and looking at their 
standard deviation over the run to estimate errors.  But I also compare 
their average to the photometry to catch gross errors (like wrong star 
identification - which I've done on occasion).

The original set used a radius of 6 pixels.  The standard deviation of 
the two reference stars over a 2 hour plus run was 3 mmag and their 
averages were very close to the photometry (I am not transforming the 
data - no time for an extra filter during the run).

The results with a 7 pixel radius were essentially identical to the 6 
pixel values.  But the 8, 9 and 10 pixel results were significantly 
different - not in the standard deviations which were still 3 mmag, but 
in the averages of the target and the reference stars.  These all 
increased (in magnitude) by about 0.3 mag.  (The three results for 8, 9 
and 10 pixels were essentially identical.)

The depth of the transit dropped by the discrepant 3 mmag but I'm 
concerned about the 0.3 mag change in the averages (which put the 
reference stars significantly off from the photometry).  For "normal" 
reporting of magnitudes to the AAVSO, how would one choose between such 
different results depending upon the aperture size used to reduce the data?

(Note:  the star images over the 2 hour plus time varied with seeing and 
wind gusts but the air mass change was very little.)

Jim Roe [ROE]