iraf.net

Dear IRAF,Can you help with this question?
I have three comparison spectra for each science object.
The spectra consist of the following:
1. a short FeAr exposure (so the red lines don't saturate our chip),
2. a long FeAr exposure (to enhance the weak blue lines, but saturate the
red lines),
3. an Ne exposureI'd clearly like to use all the relevant lines in the dispersion function
solution. Trouble is, I don't know how to do this. I have tailored the
three line lists so that only the relevant lines in each exposure are
included.How do I go about "identify"ing the lines and then fitting a dispersion
function from the entire list?Is this reasonable?Thanks for your help.Michael--
Michael De Robertis Phone: (416) 736-2100 x77761
Graduate Programme Director FAX: (416) 736-5516
Dept. Physics & Astronomy Email: mmdr@yorku.ca
York University
4700 Keele St.
Toronto, ON M3J 1P3 Canada

Hi Michael, I hope all is going well with you. We hope to see you in Tucson again
sometime soon. As for your calibration question... Here are my suggestions which I am also sending to Frank in case he has
some other suggestions. You did not say, but were your three comparison spectra taken such that
the spectral features are aligned in pixel space? 1) If so, then you could use 'imcombine' with a rejection operation in order
to create a "master" comparison spectrum. You could then use this "master"
spectrum to derive a dispersion solution based upon your relevant features. 2) If not, then I suppose you could - Derive a dispersion solution for each individual comparison exposure.
- Use 'scombine' to combine the three comparison spectra to produce a
"master" spectrum.
- If necessary, refine the dispersion solution of the "master" spectrum. Since this second solution would involve resampling of the data during
the 'scombine' phase, I am not sure of the quality of the resultant
dispersion solution. If a comparison spectrum only has "good" features at the blue or red end,
then you could try using only a subset of this spectrum for the best results
(e.g., avoid combining saturated lines with good features). Best,
Michele

Hi Michael,There is a somewhat obscure feature of IDENTIFY that is intended for
situations like yours.You would do IDENTIFY on each of your comparison spectra to define the
lines. You don't have to bother with a fit though you can just to make
sure the identifications make sense and to weed out bad lines. This
will make database files for each one. When you get to the last one or
do IDENTIFY on whatever image you will ultimately use as the comparison
reference name for DISPCOR, you add the features from all the exposures
with the ":add" command. This is like ":read" but instead of replacing
the list and reading the dispersion function, this only adds new
features to the list. This will end up creating a combined list of
features. Note that this reads the pixel positions and true
wavelengths whether or not the actual file being identified has them.
You then do a 'f' fit to fit all the lines at once for a dispersion
solution.This is a quick description. If this is not sufficient let us know.
By the way Michele's suggestion of combining the spectra before
identification is also a possibility though the combined spectrum might
become too cluttered or broaden overlapping lines.Yours,
Frank Valdes

Thanks Frank. I'll try this. Fortunately my two calibration lamps apply
to two very different parts of the spectral range and I could cobble
together a hybrid spectrum. But it's nice that one can add even more
spectra.MichaelOn Thu, 6 Jun 2002, Frank Valdes wrote:> Hi Michael,
>
> There is a somewhat obscure feature of IDENTIFY that is intended for
> situations like yours.
>
> You would do IDENTIFY on each of your comparison spectra to define the
> lines. You don't have to bother with a fit though you can just to make
> sure the identifications make sense and to weed out bad lines. This
> will make database files for each one. When you get to the last one or
> do IDENTIFY on whatever image you will ultimately use as the comparison
> reference name for DISPCOR, you add the features from all the exposures
> with the ":add" command. This is like ":read" but instead of replacing
> the list and reading the dispersion function, this only adds new
> features to the list. This will end up creating a combined list of
> features. Note that this reads the pixel positions and true
> wavelengths whether or not the actual file being identified has them.
> You then do a 'f' fit to fit all the lines at once for a dispersion
> solution.
>
> This is a quick description. If this is not sufficient let us know.
> By the way Michele's suggestion of combining the spectra before
> identification is also a possibility though the combined spectrum might
> become too cluttered or broaden overlapping lines.
>
> Yours,
> Frank Valdes
>--
Michael De Robertis Phone: (416) 736-2100 x77761
Graduate Programme Director FAX: (416) 736-5516
Dept. Physics & Astronomy Email: mmdr@yorku.ca
York University
4700 Keele St.
Toronto, ON M3J 1P3 Canada