iraf.net

Hi Frank,
Here is some info you requested: the image is 512x352, dispax=2, 34
apertures, and the apsum params are below. I have not seen the error
in an splot or implot. In fact, I can preview, but skip the orders
near where the problem has occurred. It may be due to funny values in
the data, but it was a bit puzzling why I could splot and have no
problem. There are typically wild values near the image borders. (By
the way, would it be possible to have apsum check for the existence of
the output image BEFORE starting to sum? I have often re-extracted
only to have the write fail at the end because I didn't delete the
previous attempt.) I will have to re-create such an image if you want
the data, as I have again reverted to our local software to extract
the spectra. There are various reasons why I am not using apsum, and
I intend to send a detailed report. Some has to do with the nature of
the FOE data itself, and some with software. Let me mention a couple
of things briefly.I think some of the "jaggies" seen in summed orders come from the
scattered light correction. The FOE scattered light shows a
"fishbone" pattern when displayed as an image. These correspond
directly to sawteeth in the summed spectra. They are most severe
where the orders are close together and curve across the pixel
columns, and I think may be due to the interorder minima bouncing up
and down due to sampling. IRAF seems to have more trouble with this
than BFPS (BMO Frame Processing System) does, for reasons unknown to
me.The other difficulty is more fundamental, and probably cannot be
easily solved. Very high S/N spectra typically fail to optimally
extract. I believe this is due to the local scattered light affecting
the local profile, and this depends upon the incident spectrum. Many
deep absorption lines will be mimiced in the background. I have had
consecutive short spectra of featureless hot star spectra and densely
absorbed cool star spectra, and the hot star extracts well, while the
cool one does not. My personal "solution" is to non-optimally extract
the well exposed orders, and optimally extract the lower signal orders in
spectra where it matters.I have a pile of plots and will write up a description and send it along.
Dave.--------------------------------------------------------------------------
cl> lpar apsum
input = "mr3089.019.zmin" List of input images
(output = "") List of output spectra
(sky = "") List of output sky spectra
(references = "mr3089.012.dbias") List of aperture reference images
(profiles = "mr3089.012.zmin") List of profile reference images\n
(recenter = no) Recenter reference apertures (if defined)?
(find = yes) Find apertures automatically (if none)?
(trace = yes) Trace aperture features?
(extract = yes) Extract 1D aperture sums
(skyextract = no) Output sky spectra (if background subtracting)
(background = "none") Background to subtract (none|average|fit)
(clean = yes) Detect and replace bad pixels?\n
(interactive = yes) Run task interactively?
(edit = yes) Define and edit apertures (if interactive)?
(review = yes) Review extractions and outnames (if intera
(weights = "variance") Extraction weights (profile|variance)
(naverage = 1) Number of profiles to average
(interpolator = "spline3") Type of image interpolation
(nclean = 3) Number of pixels to clean per profile per poin
(lsigma = 5.) Lower rejection threshold
(usigma = 5.) Upper rejection threshold
(v0 = 10.12) Variance intercept
(v1 = 0.0758) Variance slope
(mode = "a")--------------------------------------------------------------------------
My method has been to: Edit "fat" apertures on bias subtracted image, and save them
(to use for summing data, -2.5:2.5 from center); Run apscatter, but narrow the apertures to -1.5:1.5 so that there is
enough interorder minima to work with, and fit them with a high
order (~30) function which passes through all the minima. Run apsum on the scattered light corrected image, with "fat" apertures.