Hi Nikolay,
if switch-on ClipLowRange (and don't change any other settings) Winsorized Sigma Clipping pass cosmic ray.
This is normal behavior. When you activate one or both range clipping features, in general you must also redefine your statistical pixel rejection thresholds (e.g. sigma clipping thresholds). This happens because range clipping is applied
before statistical pixel rejection. When you enable
clip low range and/or
clip high range, the set of pixels that survive range clipping may be different for some pixel stacks, and when this happens the statistical properties of those stacks are also different.
In your example, you've found a case where a particular combination of range clipping limits and rejection thresholds is leading to significant differences in the final result. I'd like to play with this particular data set, if you don't mind, because it seems an interesting test case. A small crop covering the area in your GIF animation would be sufficient. Sorry, I am always asking you to upload images!
A different question is if range clipping should be applied
before or
after statistical pixel rejection. It seems to me that the most logical way is to apply range clipping as an initial step, because range-clipped pixels are in general marginal data (e.g. abnormally low and abnormally high pixels, or
definite outliers), and hence they shouldn't enter the statistical pixel rejection process.
Thanks for your thorough testing!
