iraf.net

I am looking in the help page for zerocombine, and I have a question about the "scale" option:scale = "none" (none|mode|median|mean|exposure)
Multiplicative image scaling to be applied. The choices are none, scale by the mode, median, or mean of the specified statistics section, or scale by the exposure time given in the image header. I am not sure I understand what this option means, and how do I decide which one is the best to use?
Could anyone please help me with this?Thanks!

I'll explain the scaling option with an example:
Say you have 3 domeflats with different exposure time and the mean values of the frames are
1000
2000
3000Now of course you want to combine these frames to create a Masterflat. So...you do not scale the data and just combine them with a median filter, but the median in this case will (most certainly for every pixel) be always the second image of your set, so the masterflat you've created will just be the second image.
What you want instead is a better option and therefore you scale your data, so by choosing mean-scaling the first image stays as it is, the second image will be divided by 2, the third by 3, so that their mean values are equal. If these scaled images are now combined with a median filter you will get a much better estimate of your pixel-to-pixel variations.
In case of flatfields, I suggest mode scaling, although with dome flats you could also chose exposure scaling, since the illumination will be more or less constant.
For Biasframes, I would chose not to scale, but this also depends on your data. If you have a significant drift during the night in the bias level and have no overscan at your hand, you might want to scale and take the small error you make into account...The point is:
You have to decide for yourself whats best for your data, but if you are not sure, don't scale the bias, darkframes (with unequal integration times) and flatfields have to be scaled.Most common scaling is:
Bias: none
Flats: mode
Darks: ExposureBut again, this depends on your data and you should give it at least some thought, because the better you reduce your data, the better your results will be.cheers

I tried the different options for the "scale". But I am not sure I understand the final output... I would imagine theat if they are identical the final resut will be a straight horizontal line exactly = zero, but for some reason that is not the case. Here are the results I got:
http://www.physics.uc.edu/~asadrs/ZerosThis is the code I used:[code:1:bc35d81d1b]## In Blanco_n1/ directory:
ccdproc images="*.imh" output="" ccdtype=" " max_cache=0 noproc- fixpix- \
overscan+ trim+ zerocor- darkcor- flatcor- illumcor- fringecor- readcor- \
scancor- readaxis="line" fixfile="" \
biassec="[3093:3132,1:718]" trimsec="[1000:3000,*]" zero="" dark="" flat="" illum="" fringe="" \
minreplace=1. scantype="shortscxan" nscan=1 interactive- function="legendre" \
order=4 sample="*" naverage=1 niterate=1 low_reject=3. high_reject=3. grow=0.cd ../cp -r Blanco_n1/ B1s/B1_none/
cp -r Blanco_n1/ B1s/B1_mode/
cp -r Blanco_n1/ B1s/B1_median/
cp -r Blanco_n1/ B1s/B1_mean/
cp -r Blanco_n1/ B1s/B1_exposure/cd B1s/B1_none/zerocombine input="zero*.fits" output="Zero_none" combine="average" reject="avsigclip" \
ccdtype=" " process- delete- scale="none" statsec="" nlow=1 nhigh=1 \
nkeep=1 mclip+ lsigma=3. hsigma=3. rdnoise="0." gain="1." \
snoise="0." pclip=-0.5 blank=1.cp Zero_none.fits ../cd ../B1_mode/zerocombine input="zero*.fits" output="Zero_mode" combine="average" reject="avsigclip" \
ccdtype=" " process- delete- scale="mode" statsec="" nlow=1 nhigh=1 \
nkeep=1 mclip+ lsigma=3. hsigma=3. rdnoise="0." gain="1." \
snoise="0." pclip=-0.5 blank=1.cp Zero_mode.fits ../cd ../B1_median/zerocombine input="zero*.fits" output="Zero_median" combine="average" reject="avsigclip" \
ccdtype=" " process- delete- scale="median" statsec="" nlow=1 nhigh=1 \
nkeep=1 mclip+ lsigma=3. hsigma=3. rdnoise="0." gain="1." \
snoise="0." pclip=-0.5 blank=1.cp Zero_median.fits ../cd ../B1_mean/zerocombine input="zero*.fits" output="Zero_mean" combine="average" reject="avsigclip" \
ccdtype=" " process- delete- scale="mean" statsec="" nlow=1 nhigh=1 \
nkeep=1 mclip+ lsigma=3. hsigma=3. rdnoise="0." gain="1." \
snoise="0." pclip=-0.5 blank=1.cp Zero_mean.fits ../cd ../B1_exposure/zerocombine input="zero*.fits" output="Zero_exposure" combine="average" reject="avsigclip" \
ccdtype=" " process- delete- scale="exposure" statsec="" nlow=1 nhigh=1 \
nkeep=1 mclip+ lsigma=3. hsigma=3. rdnoise="0." gain="1." \
snoise="0." pclip=-0.5 blank=1.cp Zero_exposure.fits ../cd ../imarith Zero_none.fits - Zero_mode.fits none_mode.fits
imarith Zero_none.fits - Zero_median.fits none_median.fits
imarith Zero_none.fits - Zero_mean.fits none_mean.fits
imarith Zero_none.fits - Zero_exposure.fits none_exposure.fits
imarith Zero_mode.fits - Zero_median.fits mode_median.fits
imarith Zero_mode.fits - Zero_mean.fits mode_mean.fits
imarith Zero_mode.fits - Zero_exposure.fits mode_exposure.fits
imarith Zero_median.fits - Zero_mean.fits median_mean.fits
imarith Zero_median.fits - Zero_exposure.fits median_exposure.fits
imarith Zero_mean.fits - Zero_exposure.fits mean_exposure.fits
ls Zero* > Zerolist
implot @Zerolist[/code:1:bc35d81d1b]

Does my question make sense?

You have to dived your results not subtract them from each other. If the result in your subtraction is zero that means, that your images are totally equal and also the scaling was equal.
But...I'm not quite sure what you want to find out with this...tell me what kind of images you have and what science you want to do with that.
Also I have seen in the headers of your image that you have an overscan, so don't worry too much about such variations.
What is the mean value of a biasframe when you subtract the overscan? If this is ~0, that means you don't have a readout bias and the subtraction of a Masterbias won't have much effect and in some cases makes it even worse.
If its unequal 0, the overscan correction takes care of your amplifier bias and a subtraction of a Masterbias is required to correct for the read-out bias.

[quote:fea0064381="AnTaR3s"]You have to dived your results not subtract them from each other. If the result in your subtraction is zero that means, that your images are totally equal and also the scaling was equal.
But...I'm not quite sure what you want to find out with this...tell me what kind of images you have and what science you want to do with that.
Also I have seen in the headers of your image that you have an overscan, so don't worry too much about such variations.
What is the mean value of a biasframe when you subtract the overscan? If this is ~0, that means you don't have a readout bias and the subtraction of a Masterbias won't have much effect and in some cases makes it even worse.
If its unequal 0, the overscan correction takes care of your amplifier bias and a subtraction of a Masterbias is required to correct for the read-out bias.[/quote:fea0064381]
What I want to find with this is: which option is best to use with my zeros. I tried all options to see if there is any significant difference between them.I have recently observed spectra of star clusters, so I am starting to apply the data reduction rules, and I am trying to get the best of my data, so I am trying all options to see the differences and choose the best...
For the question:
What is the mean value of a biasframe when you subtract the overscan?
How do I find that?
:shock:

Do imstat on your images, this will print some statistics, but let me tell you, that such minor differences won't affect your data significantly, flatfielding is a more important topic.