iraf.net

In the manual I am following it says that skyflats when combined should be scaled by "mode", but my skyflats have different exposure times, can I scale them by exposure time instead? (both exposure time AND mode are different how do I deal with such a case?)

If you have skyflats an exposure scaling does not make sense since the illumitnation changes from image to image! Scale them with the mode, thats the best choice.

So for future observations, do I need to make sure all my skyflats have the same exposure time or it doesn't matter?

When you take skyflats its important that you know your system. But as a rule of thumb you can take exposures from 5s to 30s because- <5s the shutter opening might not be negligible and therefore visible in your signal (depends on the camera)
- >30s the star-spurs become too large and too bright relative to your backgroundIn any case you have to chose your exposure time according to present illumination. If you have a 16bit detector your mean value accross the frame should be anywhere between 20000 and 40000 ADU (also depending on your camera). If its larger you risk, that some pixels are saturated, if lower your S/N is not optimal. This also depends on your system, so I would recommend to ask experienced observers with the particular telescope what they think is the best method. For our telescope here I chose values between 30000 and 40000 ADU because I know our detector is linear and at these values there are no pixels saturated.
So the exposure time is set according to your ambient light and therefore you have to chose the mode-scaling when combining the frames.

:shock: :shock: :shock: I took some exposures as long as 200s! Because it was getting dark and the signal was weak,,, do you think I shouldn't use the longer exposures when I combine my skyflats? Images Exp Mode Scale Weight
sflat1044.fits 16.0 5380.8 0.318 0.629
sflat1045.fits 30.0 482.13 3.549 0.056
sflat1046.fits 60.0 662.41 2.583 0.077
sflat1047.fits 110.0 450.88 3.795 0.053
sflat1048.fits 200.0 1578.2 1.084 0.184

OkDo you have a 16bit A/D converter? Because if you have, then the signal is definitely too small!
Of course for some systems its not possible to get many flats, because e.g. the read-out time is so large. In any case you should make as many flats as possible.
For a 14bit converter these values (except maybe the first) would still be too small, so you should consider starting earlier with making skyflats. If thats not possible you might want to switch to domeflats.
I don't know your system, so I suggest you ask someone who has some experience with the telescope!P.S.: Also there is a difference between photometry and spectroscopy. In spectroscopy you have to just look at your slit. The mode value across the complete frame is not important!Edit:
I see. You have spectroscopic data where different rules apply. The suggestions above are for direct imaging on a CCD!

When I do my skyflats (for photometry work) I generally take them 10-15 minutes after sunset and can get 25000-45000 counts on a 16bit chip in any broadband filter in around 5-10 second exposures.What I do is just set the initial exposure for a filter to give me counts near the high end (~45000) and run a series of 9 images all at that same exposure time value. By the end of the 9 images the counts are normally down near the bottom end (~25000). When I combine these 9 flats to get a master skyflat I use "median" combine with "avsigclip" and a minimum to keep of "5". This gives me good results in the flat correction so it works for me. I don't see the point in hassling to try and get the same counts across all the images when taking them at sunset since it wastes so much time when light is quickly fading away. Instead I just make sure the counts are in the right range and average them all together in the end. When you think about it this should work since the flat correction is essentially getting rid of brightness/contrast variations across the image which should have the same relative brightness to one another in any image regardless of the total average brightness of the image.

---

Don Schumacher, Jr. MS Space Sciences Florida Tech