You are here

Spectroscopy

, Updated

Spectroscopic (Telluric) Standards

Telluric standard star observations are required for all near-infrared spectroscopic observations to cancel telluric (atmospheric) absorption features in the data.  (Additional standard observations such as for flux calibration or radial velocity measurements are not part of the baseline calibration set, and are not discussed further here.)  The following is a guide to assist in selecting the most appropriate telluric standard stars.

Quick Links:


Airmass considerations - correcting for telluric absorption

When specifying standard stars, select stars whose air masses will be the same as the average airmass of the target when they are being observed. This is particularly important for observations in the L and M windows, but is also important for reducing spectra in the shorter wavelength bands. In general one should not choose standards which are at the same RA as the science target. Normally the standard will be observed just prior or just following the target, and this time interval should be taken into account, particularly if the integration on the target is a long one. As an example, if the science target spectrum will require 2 hours of time (including overheads), acceptable "before" and "after" standards would have approximately the same declination as the target but RAs approximately 1 hour less than and greater than the target. Other choices of RA and Dec also will satisfy the airmass matching criterion, but are more sensitive to the hour angle of observation.



Exposure Times for Standards

Because spectra are often obtained in better conditions than those specified in Phase2, exposure times on standard stars be set conservatively. For GNIRS one can refer to this table for guidance.


1-5um Spectral Features in Standard Stars: Which Spectral Type is Best?

So-called "telluric standards" are not standards in the classical sense at all; they are simply bright stars of known spectral type for which the intrinsic stellar features are either negligible, removable using model spectra, or easily separated from features introduced by the earth's atmosphere.  Spectral features in main sequence stars (i.e., dwarfs, luminosity class V) are present throughout the 1-5um spectral region.  The choice of the best spectral type depends on the resolution of your data and the wavelength regime of interest.  Early- to mid-A and late B dwarfs possess primarily hydrogen recombination lines, which are quite strong and highly pressure broadened.  However, these lines can be fit and removed in a fairly straightforward way, and techniques and software exist to do this, so these are the spectral types most commonly used. . Spectral type 0 and early B have very few spectral features, but are relatively rare and so may be hard to find at a good airmass match to your target.  If you are interested specifically in measuring hydrogen line profiles in your science targets, they may not be the best choice.  At G0V the hydrogen lines are considerably weaker (and many of the weaker lines seen in A dwarfs are essentially negligible); however, lines of Mg, Si, Fe and other elements have become strong.  Intrinsic lines in G2V spectra can be fit using high-resolution, high signal-to-noise solar spectra.  At early-mid F many of the hydrogen lines still are intermediate in strength and width and almost all of the lines of other atomic species are not yet very noticeable. Hence these spectral types may offer a reasonable compromise, particularly at low to moderate resolution and when the S/N of the target spectrum is low (<30, say). Later spectral types (late G, K and M) have too many intrinsic features to be generally useful for telluric corrections.  At high resolution O, B and early A dwarfs are best if the target spectrum is high S/N. In general one should look specifically at the wavelength range of interest to see where the telluric features fall, and where the stellar lines are for stars of various spectral type.  This is only a brief summary-- listed below are a few references from the literature on this subject:

A Method of Correcting Near-Infrared Spectra for Telluric Absorption, Vacca W., Cushing M. & Rayner J., 2003, PASP, 115, 389 [ADS]

Penetrating the Fog - Correcting Ground based CCD Spectroscopy, Stevenson, C.C., 1994, MNRAS, 267, 904 [ADS]

A Spectral Atlas of Hot, Luminous Stars at 2 Microns, Hanson M., Conti P. & Rieke M., 1996, ApJS, 107, 281 [ADS]

A list of hydrogen recombination line wavelengths is available.

See the NOAO Digital Library for a collection of medium and high-resolution stellar spectra in the infrared (J, H, K and L) and a high-resolution telluric spectrum from 1-5um


Magnitude limits

Magnitude limits vary with instrument and mode; higher resolution observations naturally require brighter standards.  In order not to degrade the science data during telluric division, signal-to-noise in the standard observations should be several times that of the science data whenever possible.  One should endeavour to find the brightest standards that can be used for a given instrument and mode such that e.g. S/N >100 is achievable in no more than a few minutes on-source. See the individual instrument webpages for guidelines.


Obtaining 2MASS JHK magnitudes of spectroscopic standards from the OT using the Catalog Navigator

To obtain 2MASS JHK magnitudes of your selected standard, do the following in the OT:

  • a. In the "Target Environment," enter the name of the std (e.g. HIP 23456) and hit "return," and then click on "Image.":

  • b. On the image screen (i.e., "Position Editor") click on the "Catalog" pulldown menu. First click on Image Servers and select the one you want; then click on "Catalogs" (at very top of pulldown menu) and select 2MASS.

  • c. Again on the image screen, open the "Catalogs" (Catalog Navigator) window which gives a table of all 2MASS objects. Then in the image itself, click on the std; this will highlight the std and its JHK magnitudes in the "Catalog Navigator" table.

  • d. Check to see that the JHK magnitudes and colors do not suggest a bright unresolved companion.


Hipparcos Standards

This table lists all 19,015 stars included in the Gemini telluric search utility grouped by spectral type and sorted by right ascension.

Hipparcos Telluric Standard Stars
O*V
B0V B1V B2V B3V B4V B5V B6V B7V B8V B9V
A0V A1V A2V A3V A4V A5V A6V A7V A8V A9V
F0V F1V F2V F3V F4V F5V F6V F7V F8V F9V
G0V G1V G2V G3V G4V G5V G6V G7V G8V G9V
K0V K1V K2V K3V K4V K5V K6V K7V K8V K9V

You can use the CADC Hipparcos catalogue search engine to find your own Hipparcos calibration stars. The Canadian Astronomy Data Center is operated by the Dominion Astrophysical Observatory for the National Resarch Council of Canada's Herzberg Institute of Astrophysics.


Verify your standard stars

Many stars in the Hipparcos catalog are not suitable to be used as telluric standard stars, either because of multiplicity or peculiarities in their spectra. It is the PI's responsibility to verify that the standard stars they have selected in their Phase II will meet their scientific requirements. A good way to check your star is to look it up in Simbad.


Very bright star lists

For very high resolution (Phoenix and GNIRS R=18000), this table of Telluric Reference Stars for Cerro Pachon (from NOAO), provides lists of very bright stars (limiting magnitude 3) in the southern hemisphere, grouped by local sidereal time.


Wavelength Calibration

Nearly all near-infrared spectra, except perhaps certain regions at very high resolution, contain sky emission lines which can be used for wavelength calibration.  However, at low and moderate resolution these lines are often blends and may not be easily identified.  For wavelengths < 2.5 um, measurement of arc lines for wavelength calibration is recommended.

Arc lamp line plots and lists

  • NIRI plots

  • GNIRS plots

  • Interactive tools for displaying high-res arc lines (from Keck/NIRSPEC)

  • IRAF Argon line list, including vacuum wavelengths of the infrared emission lines of argon
    from "Wavelength Standards in the Infrared", K.N. Rao et al., 1966
    [ADS] [DATA]

  • Wavelength Calibration of Near-Infrared Spectra, Hinkle, et al, 2001, PASP, 113, 548
    [ADS]

Atmospheric Transmission/Emission spectra

Sky line lists

Additional References

    • Night-sky spectral atlas of OH emission lines in the near-infrared,
      Rousselot, P., Lidman, C., Cuby, J.-G., Moreels, G. & Monnet, G., 2000, A&A 354, 1134,
      [ADS] [DATA (0.614-2.624 microns)]
    • The OH airglow spectrum: a calibration source for infrared spectrometers,
      Oliva, E. and Origlia, L., 1992, A&A, 254, 466
      [ADS] [DATA]
    • Observations of the OH airglow emission,
      Maihara, et al, 1993, PASP, 105, 940
      [ADS]
    • Non-thermal emission in the atmosphere above Mauna Kea,
      Ramsay, S.K., Mountain, C.M. & Geballe, T. R.,
      [ADS]

    • High-resolution Fourier transform spectroscopy of the Meinel system of OH,
      Abrams, et al, 1994, ApJS, 93, 351
      [ADS]
    • IRAF OH line list, including Vacuum wavelengths from the calculations of Chamberlain (< 1.20 microns)
      and laboratory observations of Hubbard and Brault using the Solar FTS (> 1.2 microns).
      [DATA]

Spectral Templates

Spectral templates (usually late type stars) are required for the analysis of kinematical data on external galaxies (or other stellar ensembles). In the near-IR, the most commonly used features are the CO overtone bands at λ > 2.3µm. Although observational and theoretical libraries exist at lower spectral resolutions (R<3000), no comprehensive set of stellar kinematic templates was available to be used with two of the configurations of Gemini NIR instruments used for stellar population kinematic studies in external galaxies - NIFS and GNIRS 111 l/mm grating (both longslit and IFU) - and all programmes using those configurations would invariably spend some science time taking a small set of stellar spectra to use as templates. This led to a constant duplication of data taking, since those targets are programme calibrations and are not made available to other users until the end of the default 18 months proprietary period.

During period 2006B at Gemini South, given the unusually poor conditions over the whole semester, and the eventual end of true "poor weather" programmes in the queue, a Director's Discretionary "poor weather" GS-2006B-DD-3 programme was carried out to provide the NIR community with a larger set of late (F7 to M3 types I, II, III and V) stellar spectra, with intermediate S/N (30-50), including the four CO overtone bands (2.24-2.42µm) at R~6000 resolution. A subset of the targets was also observed at a slightly bluer spectral range to improve usefulness for NIFS users, overlapping with the red setting on the first two CO bands.

To the original sample of 29 stars observed with GNIRS, another 31 were added from NIFS observations obtained as part of programmes GN-2006A-SV-123, GN-2006B-Q-107, GN-2007A-Q-25, and from archived public data (from programmes GN-2006A-C-11, GN-2007A-Q-45 and GN-2007A-Q-62), covering the full range 2.1 to 2.5µm at a similar resolution.

A separate R~18000 spectral template library is also available.

Detailed description of the observations, data reduction, template usage and systematic effects related to differences in resolution, the range of CO band equivalent width spanned by the templates, and further analysis in the "template mismatch" issue, can be found in "The Gemini Spectral Library of Near-IR Late-Type Stellar Templates and Its Application for Velocity Dispersion Measurements". Winge, Cláudia; Riffel, Rogemar; Storchi-Bergmann, Thaisa, 2009, ApJS, 185, 186.


Quick Links:


Library Description

GNIRS sample (at Gemini South)
Instrument configuration GNIRS IFU + 111 l/mm grating in the K band
"Red" setting λc=2.335µm (2.24-2.43µm), dλ=1.84Å/pixel
"Blue" setting λc= 2.245µm (2.15-2.33µm), dλ=1.85Å/pixel
Combined spectrum 2.15-2.43µm, rebinned to dλ = 1Å/pixel
Observed sample 29 objects in the "red" setting, from F7III to M3III
Of these, 23 objects were also observed in the "blue" setting
NIFS sample 1 (v1.5)
Instrument configuration NIFS IFU + K grating + HK filter, 3.0arcsec or KG3+ND masks
Original sampling λc=2.20 and 2.25µm (2.07-2.47µm), dλ=2.13Å/pixel
Observed sample 3 objects centred at 2.20µm, 8 centred at 2.25µm, spectral types from G8II to M5III
NIFS sample 2 (v2.0)
Instrument configuration NIFS IFU + K grating + HK filter, 3.0arcsec or KG3+ND masks
Original sampling λc=2.20 (2.04-2.43µm), dλ=2.13Å/pixel
Observed sample Spectral types from G8III to M3I


The GNIRS sample

The observed sample was selected from a list kindly provided by Greg Doppmann, compiled from the literature (mostly based in Cayrel de Strobel et al 1997), and the selection was based exclusively on observability: targets which were visible for as long as possible during the 06B semester, bright enough to provide the desired S/N on a reasonable on-source time under CC=90, IQ=ANY conditions, and having a hot (A0-A7) star close enough (and bright enough!) to be used for telluric correction. The fact that both target and telluric stars also had to have a bright (V<13mag) guide star available as well, restricted even more the choices.

The observing conditions also determined the instrument configuration: to achieve R=5900 with GNIRS in long slit mode, one would have to use the 0.30" slit - implying in very large slit losses under IQ=Any (FWHM>0.80" in K) seeing. Given the superior performance of the original GNIRS IFU in the K band (over 90% of that of the equivalent long slit mode), there was only a small loss in sensitivity by using the IFU+111 l/mm grating configuration.

The NIFS sample 1 (V1.5):

The data were obtained either as programme calibrations for GN-2006A-SV-123 and GN-2007A-Q-25, and therefore based solely in observability and brightness; or as part of another "poor weather" programme GN-2006B-Q-107, and in this case following the same rationale as the GNIRS sample (bright enough for poor conditions, with proper telluric and guide stars available). The stars observed as part of N06A-SV-123 and N07A-Q-25 used the AOWFS for guiding, while for N06B-Q-107, the AO fold was parked and guiding was done using the PWFS2.

The NIFS sample 2 (V2.0):

This extra set was derived from archived public data, originally obtained for programmes GN-2006A-C-11, GN-2007A-Q-45 and GN-2007A-Q-62, retrieved and processed by M. R. Diniz, as part of an undergraduate research project at the Universidade Federal de Santa Maria, in Brazil (with R. Riffel as supervisor).

All the GNIRS data collected under programme GS-2006B-DD-3 has been made public from the start in the Gemini Science Archive. The NIFS data were subject to the standard proprietary period. This page contains links to the table of final processed spectra. Details on the data reduction, template usage and systematic effect, can be found in the paper cited above. The community at large is welcome to download all or part of the library as needed, and users of GNIRS or NIFS are encouraged to explore its use as an alternative to requesting further observation of spectral standards with their science programmes. If the raw data are re-processed to be used in papers or publications, please use the standard Gemini acknowledgement text and the above programme IDs. For the processed data contained in this page, the authors would appreciate acknowledgement of the library use (Winge, Riffel and Storchi-Bergmann, 2009).

The data are presented in standard FITS format, and the user can select either the GNIRS red (2.24-2.42µm) or blue (2.15-2.32µm) spectral ranges at their native spectral binning; or the combined spectrum (when available), rebinned to 1 Å/pix. The NIFS spectra are also presented at native binning and rebinned to 1 Å/pix.

The authors would like to thank the former Gemini Deputy Director and Head of Science, Jean-Rene Roy, and the former Gemini South Head of Science Operations, Michael West, for the support and time allocation for our GNIRS programme. Many thanks as well to all the Gemini South observers and SSAs that so positively believed that no conditions were ever too poor to give GS-2006B-DD-3 a chance!

Current release is Version 2.0 - Additional NIFS data included.

History and release notes:

2012 Jun 25 - Version 2.0 uploaded to the Gemini Web Site - this corresponds to additional templates observed with NIFS, retrieved from the Gemini Science Archive. No changes made to the already existing spectra (GNIRS and NIFS sample 1). All files include full headers, and the data have been corrected to rest wavelength.

2009 Jan 31 - Version 1.5 uploaded to the Gemini Web Site

  • Highlights
    • NIFS data added!
    • Complete headers from the last step of reduction before extracting the spectra to 1-D added back to all spectra.
    • Continuum shape removal improved prior to combining the GNIRS settings.
    • Final spectra are corrected to rest velocity. This was done by taking a strong, isolated line at 2.2814µm and using that as a reference point to all the remaining spectra. The procedure allowed for correcting in one step the intrinsic radial velocity and any zero-point offset that could remain from the wavelength calibration.

2007 Apr 10 - Version 1.0 uploaded to the Gemini Web Site.

  • Highlights and ToDo list
    • The "blue" setting spectral range currently starts at 2.18µm. After tinkering with the telluric correction for a while, I'm still not satisfied with the result in the Brγ region.
    • The image headers are NOT complete. The original GNIRS files are MEF, so most of the information is located in the primary header unit (extension [0]), and gets lost when the spectrum is extracted. This is also in the list for the next release.
    • Final spectra are:
      • NOT flux calibrated. The continuum shape has been removed by fitting a low order polynomial. I decided to go this way rather than tackle the issue of flux calibrating the spectra for this release.
      • NOT corrected for Galactic extinction.
      • NOT corrected for heliocentric radial velocity.
    • Error vectors (variance planes) NOT included. This is admittedly important for a good evaluation of the errors in the cross-correlation results, but the variance planes are not being propagated correctly through the reduction process.


Astronomical Lines