Running TELL_CORR on X-Shooter LP spectra
=========================================

TELL_CORR interactively uses model transmission spectra for Paranal to remove telluric absorption features from 1D spectra.

The models are a function of airmass and precipitable water vapor (PWV) only.  The user adjusts these parameters, along with a velocity offset and smoothing FWHM applied to the model, in order to match the models to the absorption seen in the spectrum.

This is a beta version, so beware!

Installing:  

The following (in pro/ directory) must be in your IDL path, in addition to standard IDL libraries including the Goddard IDL Astronomy Users Library:
      tell_corr.pro
      rflam.pro
      gaussker.pro
      gxpar.pro  (Note: this is different from the GXPAR from U.Washington)

The grid of model spectra (skytable_airXXX_pwvXXX.fits) are in the telluric_models/ directory.  These models were generated individually using the SKYCALC Cerro Paranal Advanced Sky Model, version 1.3.5.  The models span a grid of air masses and PWVs, and have resolution R = 100,000.

Website: http://www.eso.org/observing/etc/bin/gen/form?INS.MODE=swspectr+INS.NAME=SKYCALC

The low-resolution extinction model for X-Shooter ('xsh_paranal_extinct_model_uvb.fits’) must also be downloaded (included here).

In tell_corr.pro, edit the variables ‘model_dir’ and ‘ext_dir’ to reflect the locations of the above.

Calling tell_corr:

   Flux spectrum only:

   IDL> uncorr_flux_file  = ‘qso_spec.fits’
   IDL> corr_flux_file    = ‘corr-qso_spec.fits’
   IDL> tell_corr,uncorr_flux_file,corr_flux_file,/newextinct

   Flux and error spectra:

   IDL> uncorr_error_file = ‘qso_err.fits’
   IDL> corr_error_file   = ‘corr-qso_err.fits’
   IDL> tell_corr,uncorr_flux_file,corr_flux_file,$
                  error_file=uncorr_error_file,$
                  correrror_file='corr_err_file’,/newextinct


Note: The /newextinct keyword must be set for LP data in order to undo the broad-band extinction correction applied to the initial extracted 1D spectra.  The high-resolution transmission models also include this extinction curve, and so this step avoids a double correction.

The user is presented with three plots: (1) the model transmission spectrum, (2) the raw flux overlaid with the smoothed and binned transmission spectrum, and (3) the corrected spectrum.  The user then adjusts the airmass, PWV, velocity offset, and smoothing FWHM applied to the model in order to best match the model to the absorption seen in the raw spectrum.

Type ‘?’ at the prompt to see a list of commands for adjusting model parameters, as well as the x- and y-limits of the display.

Tips:

- It’s often useful to zoom in on regions of a few hundred Angstroms.

- The airmass is read from the FITS header and should not need to be adjusted much. 

- The amount of O2 is absorption is a function of airmass only and does not depend on the PWV.  Since the strength of the O2 bandheads is relatively well predicted by airmass, the O2 bands are a good region for matching the velocity offset and the smoothing FWHM.

- After adjusting the velocity offset and FWHM using the O2 bands, the PWV can be set using the numerous H20 lines.

- For strong lines there will be a degeneracy between airmass/PWV and the FWHM.  Some iteration of these parameters will therefore be needed.

- The A-band (~7600 Ang) is hopeless, so don’t worry about getting it to look perfect.  Likewise for the most heavily absorbed H20 lines, particularly in the NIR.

- The parameters for the VIS and NIR arms may not be exactly the same, particularly the velocity offset (and of course the FWHM).  This may be due to a small wavelength calibration mismatch between these arms.

- When the PWV is low, different H20 bands in the NIR seem to prefer slightly different PWV values.  It doesn’t appear that this can be fixed by adjusting the FWHM, though it may relate to the resolution of the models (which affects lines that are saturated).  It’s possible that using models with R=200,000 or higher may help.  Alternatively, one could do more than one correction in the NIR and stitch the different wavelength regions together.