GESD rejects when using WBPP2.0

Roger Smith

Hi Adam,

I watched the ESD video (and many more) a couple of times, and think I understand the concept. Thanks for making it as simple as possible.

I have been using WBPP2.0 to create the master darks/flat darks/ flats, and finally make the calibrated individual frames. I am using a cooled color CMOS camera. Rejection algorithm is Auto.

Using the preset  of ESD outliers =.3

ESD significance = .05

ESD low relax =1.5

I get high reject percentages:

600S dark    5.5% high, .008%low    (65 frames)

300S dark     7.4% high, .04% low     (33 frames)

0.5S Flat darks    5.5% high, .003% low (100 frames)

Generally I like to have 1% to 2% rejects, like you target with Winsorized 

If I reduce ESD outliers to .1, then the 5%-7% will reduce. 

With ESD outliers = .3, l get low reject percentages on flats:

0.5S flat (via panel)    0.0000% high, .05% low (100 frames)

If I raise ESD outliers to .4 or .5, certainly some high side rejects will show up.

I see I can raise Flat and Darks ESD settings separately.

Please advise your thoughts on these high and low integration reject percentages?

What would make the flats data have lower rejects? Less variability, but why?

 

Thanks,

     Roger  

BlockHead

Hi Roger,

A couple of things.

1. The rejection percentage I was talking about it probably going to refer to light frames.

2. ESD does not behave the same way regular sigma reject algorithms work- and lower percentages (for rejection on light frames) is good.No need to match Winsorized Sigma Reject.

3. I have subsequently learned for ESD that the relaxation parameter affects the amount of reject at *both* the high and low end.

4. Concerning flats in general, the high S/N means the signal is much greater than the variations... so sources of noise (like readnoise and other electronic stuff... even cosmic rays) are less significant (less likely to be an "outlier.")

-the Blockhead