Dark Scaling -- Time and Temperature?

[andyschlei]

What are the limits to dark scaling?

I have a set of high quality (75 frames @ 900sec) dark and bias frames all taken at -20C. This summer and fall, I've had trouble getting to -20C for my lights.  I have assumed that the dark matching algorithms are able to correct for the variance in time and temperature.  The time difference can be significant in the case of flats which as 5-15 seconds.  The temperature difference is at most 10C (imaging at -10C instead of -20C).

Am I making a mistake in assuming I can scale on both dimensions?  I have not noticed any issues in the images, but I'd like my technique to be correct.

Thanks,

--Andy

[Juan Conejero]

Am I making a mistake in assuming I can scale on both dimensions?

Not at all. As I've said before, our dark frame adaptation routine is a purely numerical solution. It is independent on any physical acquisition conditions (temperature, exp. time, etc.). It will always converge to a solution of minimum noise in the calibrated frame.

[Lex]

Juan,

So if you always cool down your CCD as much as you can there is no need to worry for the real temperature, just see what's the temperature delta?
Is this correct or am I wrong?

Cheers

Lex

[andyschlei]

Am I making a mistake in assuming I can scale on both dimensions?

Not at all. As I've said before, our dark frame adaptation routine is a purely numerical solution. It is independent on any physical acquisition conditions (temperature, exp. time, etc.). It will always converge to a solution of minimum noise in the calibrated frame.

Juan,

Great to have definitive confirmation.  This is a tremendous time saver and since I am using a huge number of darks / baises it should be lower noise as well.

Thanks and clear skies,

--Andy

[pfile]

in my own experience with a DSLR, if the temperature is too far off, you can be left with artifacts. PI is going to minimize the total noise in the target frame but that may come at the expense of hot pixels being overcorrected. i have some Ha images of orion that ended up 'peppered' with black holes due to overscaling of the hot pixels.

i suppose maybe if you use defectmap or somesuch to handle hot pixels, you might still be okay with "extreme" scaling. but if the dark scaling factors exceed 2 or so... that's where i've seen problems.