correlated noise from BF corrections

[esheldon]

I discussed this with Daniel Gruen. He reminded me that BF corrections decorrelates the noise, that is it reverses correlated noise that is caused by BF. His comments were.

• It should decorrelate the sky noise, but do something strange to the object’s shot noise
• This means the sky noise will be more close to pure poisson
• The object shot noise should not be too strange for faint objects, mainly bright stars

As for how to proceed, there are remainiing questions:

• There will be some remaining correlated noise, but how important is that?
• Should we be trying to simulate BF on noise images, and run them through BF?

[esheldon]

more from daniel

• I would be surprised if that mattered
• the dominant noise is the background noise, and its correlation should be smaller (after the correction) than in the raw data
• and its correlation due to B/F is very small to begin with
• at 10000 background counts per pixel, there’s a correlated noise covariance of ~1 for neighboring pixels … at 100 counts, it’s 10^{-4}