[Aavso-photometry] Calculating SNR

[arne]

Richard Harvan wrote:
> Dear All,
>      I am having a difficult time determining SNR.  I suspect its operator error.  Here are the details.
>      I took 14 15-sec. images, with sky background about 5700 ADU, and variable about 25000 ADU.  The master dark was about 5100 ADU and the master flat was between 4900 and 5600 ADU.  I know the master flat is on the faint side, but this was my second try at photometry.  After processing, stacking, and summing the images, sky background was about 1.3 and the variable was 30.  I use Autostar Suite and it provides among other data, the flux,  which I understood to be the energy in the signal circle, and a count, which I understood to be the number of pixels in the signal circle.   I determined the average sky background, 1.287, multiplied it by the count, 121, and subtracted the count times sky from the flux, 287.  Taking the square root gave me the SNR of about 11.5.  
>       The check star deviations were small, on the order of 0.01 mags.  I then stacked and summed pairs of images, and did the photometry on the 7 processed images.  I obtained standard deviations of 0.017 for the variable, 0.011 and 0.010 for the check stars.  I know there is a discrepancy somewhere as a SNR of 11 and a Std Dev of 0.017 seem contradictory.  I trust the Std Dev more as I understand the process better.  What am I doing wrong for the SNR?  Has anyone using Autostar Suite obtained SNR?  I want to be able to do this since some of the data calls have a requested SNR.

So I'm confused.  If your individual images had sky at 5700ADU and target
star at 25000ADU, how did you end up with sky background 1.3 and variable 30?
My guess is that you did not normalize the flatfield before dividing.
Normalizing means to scale the flatfield so that the average pixel
value is 1.000, so that dividing this into the science frame does not
significantly change the ADU level in that frame.
Arne