You are here

Near-IR Resources

Content owned by aadamson

Gemini's near-infrared instruments are capable of imaging and spectroscopy from 1 to 5 microns (see the Instrument Overview for a brief description of each). This page provides information common to all NIR instruments, specifically targeting calibrations and data reduction resources.

Observing Conditions

(how sky conditions affect near-IR observing)

In the near-infrared (1-5 microns) a number of factors (telescope elevation / airmass, clouds, OH emission, water vapor column density, and moonlight) determine the transparency of the sky and the amplitude of fluctuations in the background radiation, both of which affect the performance of the instruments used in this wavelength region. Some of these factors are strongly wavelength dependent. The Integration Time Calculators attempt to simulate the effects of airmass, cloud cover, and water vapor, as well as image quality, and one can use the ITCs to explore the effects. The following is a rough summary of how all of these factors affect photometry and spectroscopy.


The near-infrared JHKL'M photometric filters are designed to transmit in the clearest portions of the 1.0-1.35, 1.45-1.8, 2.0-2.4, 2.9-4.1, and 4.5-5.2 micron "windows" and in some cases cover less than half of the aforementioned wavelength ranges.

In the J, H, and K photometric filters the atmospheric transmission is high and absorption by water vapor is not a significant contributor and thus there are essentially no water vapor constraints to obtaining optimal sensitivity.

Noise due to fluctuations in OH sky background emission (which occurs in various OH ro-vibrational bands in the 1.0-2.3 micron interval) is the most important source of noise in the J, H, and K photometric filters; it is most dominant in J and H. The ITCs use the nominal values for the background; but occasionally there are nights or portions of nights when these values can be exceeded by a factor of 2-3 (or can be less by that factor). OH emission varies diurnally; it is strongest in the daytime and at twilight is still typically 2-3 times higher than an hour or two later, and it doesn't increase again until just before sunrise. Thus measurements at J and H will be less sensitive shortly after the end of twilight than an hour or two later. Note also that OH emission roughly scales with airmass so the noise due to background fluctuation is higher at higher airmass.

In the L' and M filters absorption by water vapor is more significant, but still generally less important than other atmospheric constituents such as CH4, CO2, CO, and O3. Overall the atmospheric transmission is lower than in the JHK filters. It is recommended that photometric measurements in the L' and M filters avoid the highest airmasses and water vapor columns (i.e., do not specify WV "any").

Moonlight (which is not included in the ITC) can be a non-negligible source of sky background in the J band; it is much less of a problem in the H and K bands, although if the moon is very close to a faint astronomical target the performance in those bands can also be affected.


Most of the factors discussed above also affect spectroscopic performance, but the effect of each varies from band to band, and indeed from wavelength to wavelength (e.g., a sought-after emission line may or may not correspond in wavelength to a strong telluric absorption line). There also are some instances where one can characterize one part of a window as affected in a different ways than another part. For example, the 2.85-3.25μm region contains numerous strong lines of H2O, but beyond 3.25μm the both density of H2O lines and their strengths decrease markedly. Thus an investigator whose science resides in the 2.85-3.25μm region should request drier conditions than one who is interested only in 3.25-4.15μm.

The above issues mean that the investigator should use the ITCs to check the transmission and background at specific wavelengths or wavelength intervals of interest, and should vary conditions such as airmass and water vapor to see their effects.

Accurate removal of telluric lines by ratioing the spectrum of the science target by that of a calibration star is generally achievable if the airmass match between the two objects is good. However it in regions where lines of telluric H2O are important, it can fail, because unlike other important absorbing species, the water column varies temporally as well as well as with airmass. Investigators whose science lies at wavelength affected by H2O lines may wish to specify, possibly by using some science time, that calibration stars are observed more frequently than normal, in order to mitigate this problem.

Baseline Calibrations

A set of baseline calibrations will be taken for all Gemini queue observations in order to ensure the long-term utility of the data in the archive. This baseline calibration set varies from mode to mode, but includes all the data deemed necessary to produce scientifically useful data.   PIs are free to request additional calibrations beyond the baseline set, however, all calibrations, baseline or additional, must be specified by the PI in the Phase II definition.  The difference between baseline and requested calibrations is that baseline calibrations are not charged to the program, while any additional calibrations are charged to the program.  The table below describes the calibrations for each wavelength regime and delineates which are included in the baseline set. Instrument-specific variations are noted.

Baseline calibration data have no proprietary period and are publicly available in the archive as soon as they are uploaded.  These observations may be executed from a shared "Basecal" program (for shared configurations) or from within each science program. Time used to obtain baseline calibrations between nautical evening twilight and nautical morning twilight is charged to the partner country associated with the program or programs (i.e., "Nighttime Partner Calibration"). Time between morning and evening twilight is not charged (i.e., "Daytime Calibration"). Any time required for additional calibrations requested by the PI is charged to the program (i.e., "Nighttime Program Calibration"). To ensure proper time accounting, please set the Observation Class correctly on all calibration observations. Examples of baseline calibration observations are available in the OT libraries for each instrument.

Imaging 1-2.5 µm Imaging >2.5 µm Spectroscopy 1-2.5 µm Spectroscopy >2.5µm
Bad pixel mask Can be derived nightly from GCAL flats and short darks; a "canned" BPM may be available in the instrument IRAF package.
GCAL Flat Fields GCAL imaging flats specified in the Phase II will be taken the morning after the science observations; these should include shutter-open and shutter-closed exposures to allow for correction of thermal emission, dark current, and hot pixels. GCAL cannot be used for L and M-band imaging due to saturation of the array. GCAL spectroscopic flats1 will be taken either with the science program at night, or the following morning. See the OT library for each instrument for examples and more details.
Sky Flat Fields Not part of the baseline calibration set. At L and M sky flats are the only flats available and must be derived from the program data; at shorter wavelengths sky flats may be substituted for GCAL flats. In some cases dark current stability can limit the accuracy of the sky flats, so you may wish to specify dark frames.2 Not applicable
Darks If desired, dark frames for science exposures can be defined and will be taken as daytime calibrations (not during the night).  As of semester 2024B, programs using GNIRS for keyhole imaging will have dark sequences defined by default as part of the suite of baseline calibrations. For all other NIR instruments, as well as for GNIRS spectroscopy, it is expected that most PIs will use "sky" frames to remove dark current (in addition to the removal of the sky itself). Note that short darks (~1s) are taken daily for all instruments to assess readnoise and identify bad pixels.
Wavelength calibration Not applicable Ar and/or Xe arc lamp measurements should be specified in the science observation for each spectral setting.  In some cases,3 a corresponding "lamp off" exposure should be included.
For GNIRS in it highest resolution mode, in some cases the limited spectral coverage does not in include enough arc lines for an accurate calibration; sky lines in the science and/or telluric standard exposures are used instead.
Sky lines present in the science data are used to determine the wavelength calibration.
Telluric standard Not applicable We will observe one telluric standard per science target for every 1.5 hours of clock time on the science target. Continuous observations of the science target may be up to 3 hours in length, bracketed by two telluric standards. We will observe one telluric standard per science target for every 1 hour of clock time on the science target. Continuous observations of the science target may also be up to 2 hours in length, bracketed by two telluric standards.
PIs should include two standards in every observation group: one suitable for observation before the science target, and one suitable for after the science target.  See our telluric standards page for more information.
Flux standard We will observe one photometric standard for every 2 hours of clock time on the science target. The PI should supply one standard before and one after the science observation in order to give a good airmass match.   See our photometric standards page for more information. Photometric accuracy is limited to ~10% by the uncertainty in the (unmeasured) atmospheric extinction. Not part of the baseline calibration set 4
Atmospheric extinction Not part of the baseline calibration set.
Point Spread Function Not part of the baseline calibration set.
World coordinate system   Written in the header of all images. The WCS is accurate to approximately one arcsec.
Spatial Rectification mask (i.e., for S-distortion correction) Not applicable GNIRS:for cross-dispersed data an appropriate pinhole mask spectrum should be defined with the flatfields; for long slit, a pinhole can be specified if desired.
NIFS: a "Ronchi" calibration mask spectrum is provided for each wavelength setting.
Focal plane mask image Not applicable The slit or IFU will be imaged without the disperser as part of the acquisition sequence.
Radial Velocity or Line index Standards Not applicable Not part of the baseline calibration set.



1 Where necessary, flats taken with the calibration unit make use of shutter open and closed images to allow correction of dark current and hot pixels. NIRI spectroscopic flats typically only include shutter-open exposures, as the contribution from the background is negligible, and the dark current can be removed using daytime darks. NIFS daytime calibration templates include "lamp off" flats to remove dark current.

2 The calibration unit produces a beam that matches the telescope pupil very well except that it contains no central obscuration. The light path between GCAL and the instrument excludes the primary and secondary mirrors as well. The GCAL illumination is therefore subtly different from that of sky flats.  Sky flats can be constructed from data that sees exactly the same pupil.  For narrow band filters in the 1-2.5 microns region dark current subtraction can be important for making sky flats and variations of dark current over periods of a few hours can limit their accuracy. Therefore, we recommend that GCAL flats be obtained (in the daytime). If for some reason sky flats are desired for 1-2.5 micron narrow band imaging, then dark current images should be taken at the same time as the data to be used to construct sky flats (note that the program will be charged for this time). For 3-5 micron imaging exposure times are very short, dark current is a miniscule effect compared to the background, and it is not necessary to obtain darks during nighttime.

GCAL imaging flats are reproducible from night to night to about 0.3% i.e. the sensitivity of a given pixel varies by 0.3% over many nights as measured by GCAL. Obviously, for a star that subtends many pixels, the photometric accuracy will be approximately 0.3% divided by the square root of the number of pixels.

3 Darks are taken for NIFS arcs to remove hot pixels which would complicate the wavelength calibration process.

4 Flux calibration is inaccurate because of slit losses, which vary somewhat with wavelength (due to differential refraction and the wavelength dependence of seeing). For this reason, spectrophotometric standards are not done as part of the standard baseline calibration. The accuracy for which the relative flux density of a telluric standard is known is probably better than 10% for a given atmospheric window, 10-20% for the full range of a cross-dispersed spectrum. If the project requires more accurate flux calibration than this, the observer has two choices:

  • Ensure that photometry is already available for the science targets, and normalize the spectra.
  • Request time with photometric conditions for additional observations of the target and a flux calibrator with a wide slit to provide the normalization.