Hi Jeff,
I believe that Biases, for an 'ideal match', should be temperature matched to Darks and Lights. That will at least 'statistically eliminate' the temperature variable - because, after all, you have kept it 'constant'.
When it comes to Flats (a different kind of 'Lights', in reality), then I would apply the same logic. Flats, Darks and Biases all acquired at the same temperature AS EACH OTHER.
Of course - if you have a TEC camera, then it is sensible to figure out what is a SUSTAINABLE temperature for ALL of your data acquisition - and don't push this temperature setting down too low. For example, on my new QHY10 CCD, I have run trials at -15C, and using exposure times up to 30mins. It seems that -15C is an easily achievable CONSTANT temperature, even working inside the house at an ambient temperature of +20C - monitoring the TEC system, I was able to observe that the PWM drive signal never exceeded 60% of maximum, so I would have been able to handle even higher ambient temperatures, if needed (although HIGHLY unlikely here in the UK!!).
But, I would also have to consider a very 'cold' observing session - where the camera might be running in an AMBIENT environment of -15C, requiring the TEC system to run (effectively) at 0% - in other words with no real control whatsoever. Not a good mode of operation - but easily avoided because 'I' won't be out if it is THAT cold
So, in my case, I feel ready to work at -15C, and will acquire Darks and Biases at -15C. That means I will also need to acquire Lights and Flats at -15C as well. So far, no problems.
Now, my typical exposure times for Lights will (hopefully) be between 1 and 5 minutes (this may get longer if my new OAG system eliminates my current 'flexure' problems), and I tend to just expsose in 'multiples' of 60 seconds - because I don't know of any reason to do otherwise. So, what I can do (because my TEC runs so well) is simply acquire a Dark frames library of 30-odd Darks each, for each exposure time I am likely to work with. And I have already done that for 60 through to 300s, 30 frames each at -15C, with another 30 Biases acquired in between each Dark frame expsoure session. Total of 300 frames. Took about ten hours. I slept most of that, and then went to work. Easy job!!!
So, now I have 5 MasterDarks, and a 'superMasterBias' created from 150 acquired Biases - all of which appeared (statistically) 'identical', throughout the ten-hour acquisition period. Next task is to 'prove to myself' that - if necessary - temperature-equivalent Darks can be 'scaled' for exposure, just in case I decide to acquire Lights at some intermediate exposure time.
For example, I will be looking to see if a 'scaled 2s MasterDark' is statistically equivalent to an 'actual 2s MasterDark'. If this IS the case, then I don't need to go out of my way to also acquire FlatDarks, at 2 second exposure times, to 'match' my 2s Flats (assuming that 2s exposures are what I will need to achieve 50% ADU levels with the new camera and my existing LightBox).
I know that there has been a lot of discussion regarding ImageCalibration, that suggests that 'all' Darks can be fed into the calibration process - even if they were NOT 'temperature correlated' to the Lights (or Flats) that they are being used to calibrate. Whilst this may well be the case (becasue of the inherent 'power' of the IC process in PI), I see no real reason 'for' adopting this approach, when - assuming a well-controlled TEC Camera - there is no reason NOT to be using temperature-correlated exposures in the first place.
And your last question, regarding the possible range of temperature difference that could be tolerated by the IC process - I hope to tackle this, by experiment, over the next few days - by simply recording Darks and Biases at a 'fixed' exposure time (probably 120s, for this experiment) and looking at how different the resultant Bias-Calibrated MasterDarks actually are. By varying the TEC temperature plus or minus a degree at a time, I will be able to see what kind of effect this actually has on the data.
Of course, if the sky clears up, and I can actually manage to get through the snow to the observatory, and can get the roof to open, and find that there isn't an eight-foot snow drift inside, I may actually just end up imaging the heavens - in which case all these experiments will have to wait until the next cloudy night!!
