iraf.net

Hi, everyone! My question addressed to people who use Iraf for processing of spectra obtained with an image slicer. Our slicer split image into 3 slices, so 2 slices shifted relatively to central one approximately on 3 pixels. What is method you are using to process slicer spectra? At present we use set of IDL routines Reduce, written by N. Piskunov and J. Valenti. They use cross-correlational analysis for merging of slices. Is there any task which allow me to perfom such operation? And finally, what is an optimal algorithm (in your opinion) for processing spectra obtained with slicers. Thanks.

---

Cheers, Eugene.

Hi Eugene,This can be done using the standard twodspec tasks, with a bit of work. For our slicer IFU data (gnirs & nifs), we have some CL scripts in the gemini IRAF package, which are wrappers around the twodspec stuff. First, we cut out the image sections roughly corresponding to image slices, so each slice is a separate image or FITS extension. Then we run "identify"/"reidentify" on each row of each slice from an arc, to create a wavelength solution database file for each one. You may also need to do the same thing in the spatial dimension, to calibrate the spatial centre of each slice as a function of wavelength. If you are only observing stars, you could try using the stellar spectrum directly as a reference peak for identify; alternatively, you could extract each slice spectrum directly to 1D before combining them, in which case you wouldn't need to worry about calibrating the spatial dimension. If you are looking at extended sources, you can use a special calibration mask or a peak generated artificially from a flat field (see "gemini.gnirs.nfflt2pin") as the spatial reference for identify. Once you've done the identify step(s), you can run "fitcoords" to get a set of spectra calibrated in wavelength and the dimension along the slits. You then run "transform" to resample all the slices onto a linear co-ordinate system. To get all the slices onto the same grid, I think you need to specify the output co-ordinates manually for the second slice onwards. It's not too difficult to write a script that runs the first slice through "transform" and then reads CRVAL1 etc. from the resulting header and specifies the same WCS values when running transform on subsequent slices. After you've done all this, you can just co-add the slices as normal or stack them into a datacube in the 3rd dimension. If you'd like to take a look at our code for doing this, you can refer to "nswavelength", "nssdist" & "nstransform" in the gemini.gnirs package, but unfortunately there is a lot of gemini-specific error checking in those tasks that makes them quite hard to read.So, in summary, you can calibrate the slices much like individual long-slit spectra and then set the "transform" parameters so that you end up with them all on the same grid for combining.Does that help?Cheers,James.

Hi James! I'm very grateful for such comprehensive explanation.

---

Cheers, Eugene.