Question about ADU levels of Calibration Frames

[Adrien]

Hello
I have not had much success using "automatic routines" for image calibration. Its probably "operator error" So I am starting from the beginning, and am running Image integration and calibration, manually to identify where the process is going wrong.
I have a QSI 683, and the results I am getting so far are: All results are MEAN ADU

RAW Bias = 269.1
Superbias = 267.3
RAW Dark = 288.6
Masterdark = 22.34

Programmed Flat ADU level in CCDAP is 37,000 ADU

Flat = 37210
Master Flat = 36717

Example one LUM Light frame:
RAW Light = 1749
Calibrated = 1703.8

I think most of these look good, but should the Bias (267.3) not be subtracted from the light frame of 1749 to give 1481?

Appreciate any assistance.

Thank you
Adrien

[pfile]

probably the division of the flat has changed the mean value... i guess you could disable the flat and see what you get.

rob

[Adrien]

Hi Rob
When I disable Flats in "Image Calibration", the result with the same light frame is 1460 Mean ADU. That is better, but surely this is not the recommended way of calibrating frames?

Thank you
Adrien

[pfile]

well it's not that you should disable the flats, but your question was why the mean of the calibrated frame was not similar to the straight subtraction of the bias/dark from the light. what i was trying to say is that once you divide by the flat, you've left the realm of addition and subtraction and so the mean value can and will shift away from the result of operations that don't scale the data.

also by default PI will scale the dark ("optimize") so the mean value may not be what you expect from straight subtractions when the flat is omitted.

rob

[chris.bailey]

The light will have the Bias and Dark subtracted so 1460 is about right (1749-269-22). The Flats are rescaled before the division so its much harder to do simple maths to sense check the result.

If I rescale one of my master flats and apply statistics to that I get a mean of 0.85 so your end result looks "about" right.

Chris

[Adrien]

Hello Chris and Rob
I am not being "picky", but, on a monthly basis, I go to on trips to the desert, to take astro pictures, and I must know that my data is good.
Rob, my point is that when I include flats in the calibration process, the results are not 1460 but 1703.
Should I be concerned? I know every ADU counts, but I cant see why there is a difference of 242 ADU.
Thank you
Adrien

[pfile]

a master flat is applied to a light by dividing it into the light.

if you divide a number by anything other than 1, you will get a different number out of the division operation.

given that, i am not sure why you are concerned. the mean value of the all the pixels in an image is going to get larger as the result of the flattening is to brighten the vignetted areas.

rob