iraf.net

Hello,I am running noao.twodspec.apall to extract spectra from
my data (stellar) and I have noticed the following: 1. Sometimes while extracting spectra, an error message
"Warning: Polynomial separation too large" appears and
no spectra is extracted.2. The "clean" option doesn't seem to be doing it's job:
cosmic rays show up in the background Spectra and
therefore an "absorption" is seen in the stellar spectra. The data are Echelle spectra at 5890.2 A. The questions
I have are: what does the error message mean, what causes
the error and what can I do? I believe the apertures are
good. Should I remove (somehow) the cosmic rays before
running apall for the "clean" option to work? The cosmic
rays are typically 5-10 pixels in diameter, and I'm
starting to suspect if this is too wide to be considered
"bad pixels".Here are some of the parameters I used:(b_funct= chebyshev) Background function
(b_order= 2) Background function order
(b_naver= -100) Background average or median
(b_niter= 5) Background rejection iterations
(b_low_r= 3.) Background lower rejection sigma
(b_high_= 1.5) Background upper rejection sigma
(b_grow = 0.) Background rejection growing radius(llimit = INDEF) Lower aperture limit relative to center
(ulimit = INDEF) Upper aperture limit relative to center
(ylevel = 0.1) Fraction of peak or intensity for automatic widt
(peak = yes) Is ylevel a fraction of the peak?
(bkg = yes) Subtract background in automatic width?
(r_grow = 0.) Grow limits by this factor
(avglimi= no) Average limits over all apertures?(backgro= fit) Background to subtract
(skybox = 6) Box car smoothing length for sky
(weights= variance) Extraction weights (none|variance)
(pfit = fit2d) Profile fitting type (fit1d|fit2d)
(clean = yes) Detect and replace bad pixels?
(saturat= INDEF) Saturation level
(readnoi= RDNOISE) Read out noise sigma (photons)
(gain = GAIN) Photon gain (photons/data number)
(lsigma = 3.) Lower rejection threshold
(usigma = 3.) Upper rejection threshold
(nsubaps= 1) Number of subapertures per aperture
Thank you for your help,Kaori Nishikida (kaori@astro.psu.edu)p.s. I'm running Version 2.10.4

From: kaori@astro.psu.edu (Kaori Nishikida)
Subject: "apall" help
Date: 2 Jun 1997 14:33:52 -0700Hello,I am running noao.twodspec.apall to extract spectra from
my data (stellar) and I have noticed the following: 1. Sometimes while extracting spectra, an error message
"Warning: Polynomial separation too large" appears and
no spectra is extracted.2. The "clean" option doesn't seem to be doing it's job:
cosmic rays show up in the background Spectra and
therefore an "absorption" is seen in the stellar spectra. The data are Echelle spectra at 5890.2 A. The questions
I have are: what does the error message mean, what causes
the error and what can I do? I believe the apertures are
good. Should I remove (somehow) the cosmic rays before
running apall for the "clean" option to work? The cosmic
rays are typically 5-10 pixels in diameter, and I'm
starting to suspect if this is too wide to be considered
"bad pixels".Here are some of the parameters I used:(b_funct= chebyshev) Background function
(b_order= 2) Background function order
(b_naver= -100) Background average or median
(b_niter= 5) Background rejection iterations
(b_low_r= 3.) Background lower rejection sigma
(b_high_= 1.5) Background upper rejection sigma
(b_grow = 0.) Background rejection growing radius(llimit = INDEF) Lower aperture limit relative to center
(ulimit = INDEF) Upper aperture limit relative to center
(ylevel = 0.1) Fraction of peak or intensity for automatic widt
(peak = yes) Is ylevel a fraction of the peak?
(bkg = yes) Subtract background in automatic width?
(r_grow = 0.) Grow limits by this factor
(avglimi= no) Average limits over all apertures?(backgro= fit) Background to subtract
(skybox = 6) Box car smoothing length for sky
(weights= variance) Extraction weights (none|variance)
(pfit = fit2d) Profile fitting type (fit1d|fit2d)
(clean = yes) Detect and replace bad pixels?
(saturat= INDEF) Saturation level
(readnoi= RDNOISE) Read out noise sigma (photons)
(gain = GAIN) Photon gain (photons/data number)
(lsigma = 3.) Lower rejection threshold
(usigma = 3.) Upper rejection threshold
(nsubaps= 1) Number of subapertures per aperture
Thank you for your help,Kaori Nishikida (kaori@astro.psu.edu)p.s. I'm running Version 2.10.4>From valdes Tue Jun 3 10:53:25 1997
Date: Tue, 3 Jun 97 10:53:21 MST
From: valdes (Frank Valdes)
To: kaori@astro.psu.edu
Subject: Re: "apall" helpHello Kaori,If you read "phelp approfiles" you will see that there are two
algorithms for modeling the profiles. The modeling is needed for
both variance weighting and cosmic ray identifications. The more
complex model used when the profiles are strongly tilted with respect
to the image lines or columns does some 2D polynomial fitting where
the polynomial separation across the dispersion must be less than
one pixel. The default values have been set to work in most cases
but something about the tilt of your spectra, the size of the apertures,
and the parameter settings is causing a integer roundoff that violates
the requirement. If you do cl> =apall1.polysep
0.95
cl> apall1.polysep=0.9

you can check and reset the parameter that is being used. By making
it a bit smaller the error should go away.Yes the cleaning option only works well when the cosmic rays are
quite small. The cleanning is only done by comparing the modeled
profile across the dispersion axis to the data at each line or
column. So if the width of the cosmic ray compared to the profile
width is too large then it will not find a cosmic ray. It
may be that you cannot get a good result. You might try using the
alternate model fitting by setting apall.pfit=fit1d.The cleanning also does not work well if the cosmic rays are very
strong compared to the spectrum profile. The way to help the
algorithm in this case is to find a data value that is just above
the highest value for real spectrum signal and set the "saturation"
parameter to that value. Then any pixel above this value will automatically
be excluded and strong cosmic rays will be thrown out. The weaker
cosmic rays will still need to be found by the model profile
comparison to the data. You can also reduce the reject thresholds
though you will reach a point where too much real data will be thrown
out. Oh, the comparison between the data and the model profile is,
of course, based on the expected noise model. So if the readnoise
and gain are set wrong and the predicted noise is higher than it
really is then cosmic rays will also not be found with as much
sensitivity as is possible.I hope this helps in the challenging problem of cosmic ray cleanning.Cheers,
Frank Valdes