The observations were done using the Integral Field Unit (IFU) in the Gemini Near-Infrared Spectrograph (GNIRS), with the grating 110.5 l/mm, yielding a resolving power of R~6000 (FWHM=3.4Å at 2.293μm).  The list of observations is given below, where for each spectral setting we present the date(s), the total exposure time on source and the hot star used to correct for the telluric lines.  Observations were dominated by the instrumental overheads: for example, in the "blue" observation of  HD39425, the 36sec of on-source time required 4.85 minutes of clock time (begining to end of observing sequence, not including slew and target acquisition).

The standard group of observations included a science target, one or two telluric stars (giving better airmass match if observed before or after the target, or if observed before or after the target transited), a set of calibrations comprising three arcs and a set of 10 GCAL flats. Calibrations (arcs and flats) were usually observed right after the science target, or after a set of targets was observed, but before the grating was moved to another configuration.

Observing sequences were defined as several (2-5) repeats of ABBA sequences, with an p=4" offset between A and B positions. This translates to a offset perpendicular to the  long axis of the IFU field-of-view, large enough to move from a centred object completelly off to sky (although in some of the cases where the seeing was really poor it was still possible to detect the wings of the PSF in the B position).  On-target efficiency with this setup gets reduced by 50%, but it avoids the problem of overlapping PSF wings due to the small size of the IFU if trying to dither on source.

Exposure times were calculated using the GNIRS ITC for two cases: (a) the maximum exposure time that would not saturate a single exposure (1 coadd) under IQ=70%, CC=50% conditions; and (b) the integration time per exposure required to obtain the desired signal to noise under IQ=Any, CC=90% conditions. A large number of coadds (rather than a longer integration time per exposure) was used to go from (a) to (b), thus avoiding the risk of saturation if observations were carried out under variable conditions (for example, clear patches between clouds).  The same procedure was used to define  the telluric standard observations.

The GNIRS data frames as delivered from the Gemini Science Archive are in the standard Gemini MEF (Multi-Extension Format), where the primary header unit (PHU, extension [0]) includes all header information from telescope, environmental monitoring system and instrument; and the data extension [1] contains the pixel values.

Data reduction was performed using the tasks in the gemini.gnirs IRAF package, the released Version 1.9, of July 28, 2006, and comprised the following steps:

Calibrations - Flats and arcs

1. nsprepare: reformats the files to add the IFU Mask Definition File and applies the linearity correction to the data. The resulting file contains the PHU, one binary table extension with the MDF and one data extension [SCI,1] with the actual pixel values.
   
2. nsreduce: cuts each of the 21 IFU slices according to the MDF inserted above to a separate SCI extension. No dark correction is applied to either flats or arcs.
   
3. nsflat: combine the ten frames by extension, using ccdclipping for rejection and normalizing by the median of the illuminated area in each slice (as defined by the MDF). In average, the processing resulted in a S/N ~200-300 for each extension, with exception of slices 1 and 21 (which are partly vignetted) and slice 13 (which is damaged).
4. used gemcombine to average the three processed arc frames to improve visibility of faint lines.
5. nswavelength: obtain wavelength solution from combined arc. Using the Ar lamp, there are four lines in the "red" setting, and six in the "blue" setting. A low order polynomial (legendre order=3) was used, with residuals of the order of 0.15Å or smaller.
   

Science data and telluric stars

1. nsprepare: reformats the files to add the IFU Mask Definition File and applies the linearity correction to the data. The resulting file contains the PHU, one binary table extension with the MDF and one data extension [SCI,1]
2. nsreduce: cuts each of the 21 IFU slices according to the MDF inserted above to a separate SCI extension. Subtract adjacent pairs of object-sky frames and divide by the flatfield.
3. nsstack: since we had only one position with actual data (the B position was blank sky), and the target objects were bright point sources observed under poor seeing conditions, we simply stacked all A  positions without any effort to improve alignment of the individual object frames by shifting according to the offsets registered in the headers. In most cases all frames were within 0.3arcsec tolerance (according to the offsets registered in the headers), but there were a few observations where drifts of up to 0.8arcsec were seen (usually due to clouds or very poor seeing affecting guiding performance).
4. nstransform: applied the wavelength transformation to the stacked frame
5. nsextract: interactively extracted the spectrum from each slice, in order to exclude those with very low signal (the targets was not always well centred), the two edge slices and the damaged slice when the spectrum happened to fall within the damaged region. The output from this task is still a MEF file, with each SCI extension containing a 1D spectrum.
   
6. used a simple cl script wrapped around specred.scombine to combine the valid spectra obtained in step 5. With this step, a single 1D standard FITS spectrum is created, but most of the information contained in the header of the MEF files is lost (scombine propagates the header of the first extension included in the combining list).
   
7. finally, for the science data, applied the telluric correction using the standard specred.telluric task.
8. combined the telluric-corrected spectra for those targets observed more than once.
   

One additional step was applied to the data presented here, which was to remove the continnum shape by fitting a low order polynomial to the final telluric corrected spectrum.

Last update 2007 April 09 Claudia Winge