Hello! I’ve read that the best way to calibrate frames to be stacked is to create a unique dark for each frame. That is, shoot a number of darks for each science frame, apply that master dark to the individual science frame and then stack the calibrated frames.
That seems unwieldy.
It seems more common to create a single master dark and apply that to each individual science frame, or to apply it to the stacked science image.
I'm curious - how would the error range compare with each technique?
Best regards.
I’ve not done an analysis of both approaches. However, having watched the Arne videos on how to process variable star data, he takes darks (while it’s dark) then uses these darks for an extended period (like a month) to calibrate his frames. The one thing you don’t want to do is add noise to the data processing. That’s why it’s important to take lots of darks, bias, flats, etc; you need to 'beat down" the noise. The one calibration type required on a daily basis are flats. It’s amazing how quickly dust can accumulate and shift in an environment like the desert southwest.
It’s interesting to look into this. On page 50 of The Handbook of Astronomical Image Processing by Richard Berry and James Burnell (I’ve got the Second English Edition, Fourth Printing November 2011), it is stated “… in general, subtracting a dark frame increases the random noise level. However, dark subtraction also removes hot pixels, dark-noise patterning, and the bias signal. As a result, dark subtracted images are cleaner and look better than raw images.”
In this section of the book (Chapter 2, section 4.3 Signal and Noise in a Dark-Subtracted Image) there is a mathematical example relevant to the above quote.
Yes, subtracting a dark will add random noise. The purpose of a dark is to reduce systematic noise which usually dwarfs random noise; think of a dark a a low pass filter that eliminates spikes in the frame. As you said, dark subtracted images are cleaner and look better than raw images. I attribute that qualitative assessment to reduced noise.
Just curious - has there been a study on how the different dark application techniques for stacking affect noise. For example, a master dark to each sub-science flat? A master dark to the stacked science flat? A new unique master dark for each sub-science flat? Best regards.