[Aavso-photometry] Flat Stuff

[Michael Koppelman]

These seems to ignore the fact that the low S/N frames are going to be 
weighted very, very low vs. the high S/N frames, doesn't it? The 
fractional uncertainty of the high S/N frames is going to be an order 
of magnitude less than the fractional uncertainty of the low S/N 
frames.

Michael


On Jan 18, 2005, at 12:55 PM, Chuck Pullen wrote:

> Michael - I used to use frames with very different median counts on 
> them, and let them be scaled.  However, Arne suggested that could 
> introduce bias to a process that is supposed to introduce uniformity.  
> As I understand it, the pixel to pixel variation with low S/N flats is 
> the intrinsic variability **plus** the statistical light noise that 
> comes from any low S/N frame.  While you can scale that up to a median 
> of 1/2 well depth, you are simply multiplying those individual low 
> pixel values **plus** the high noise by some scaler factor.  So, you 
> actually get more noise, not less, then using frames that are all well 
> exposed to begin with.