You are in: Instruments > GMOS > Performance and Use > Observing Strategies |
![]() |
GMOS Observing Strategies |
The recommended detector read-out configuration for all GMOS observations is
It is recommended to use the OIWFS for guiding for all GMOS observations. The guide star needs to have a V magnitude between 9.5 mag and 16 mag, see the OIWFS page for details on the required magnitude as a function of observing conditions. The OIWFS provides tip/tilt guiding, focus and astigmatism corrections. When choosing a guide star, keep in mind that a brighter guide star gives better guiding performance, and therefore better image quality. You may be able to use a brighter guide star by changing the position angle for a given observation.
There is significant flexure between GMOS and the PWFSs. Thus, it is not recommended to use the PWFSs for guiding until the flexure has been resolved.
The gaps between the three detectors in GMOS cause gaps in the imaging field about 2.7 arcsec wide, see the data examples. If continuous coverage of your field is needed, you will need to obtain multiple exposures and dither between the exposures. A minimum dither step of 5 arcsec in the X-direction (p-direction on the OT) is recommended.
For unbinned observations the pixel scale is 0.0727 arcsec/pixel (GMOS-N) and 0.0730 arcsec/pixel (GMOS-S). If you are requesting imaging in image quality 70-percentile or worse, consider binning the CCDs 2x2 giving an effective pixel scale of 0.1454 arcsec/pixel (GMOS-N) and 0.146 arcsec/pixel (GMOS-S). This cuts the overheads from the readout from 150sec to 55sec per frame.
The GMOS North CCDs have strong fringing in the z'-filter, and weak but still visible fringing in the i'-filter. The peak-to-peak amplitude is about 5% for the z'-filter. The GMOS South CCDs have strong fringing in both the i' and z' filters, the peak-to-peak amplitude is about 50% for the z'-filter. For sparsely populated fields, fringe frames can be constructed from the science images. For crowded fields or extended objects, you may consider obtaining addition sky observations for construction of fringe frames. Example fringe frames are available (GN, GS). To construct a reasonably good fringe frame you will need a minimum of six dithered science exposures.
Are the Baseline Calibrations sufficient for your program? If you need photometry to better than 5%, you will need to add standard stars to your observations to determine extinction.
The gaps between the three detectors in GMOS cause gaps in the spectral coverage, see the data examples. The gaps are roughly 37 unbinned pixels wide. The size of the gaps in wavelength space depends on the grating used, but is typically a few nanometers. If continuous spectral coverage is essential for your program, consider using two configurations of the grating with central wavelengths 3-5 nm different.
The GMOS long-slits have bridges that cause gaps in the spatial coverage, see the data examples. The bridges are 3.2 arcsec wide. If continuous spatial coverage of your long-slit observations is needed, consider dithering in the Y-direction (q-direction on the OT). A minimum dither step of 5 arcsec is recommended.
If you are requesting long-slit observations in image quality 70-percentile or worse, consider binning the CCDs by two in the spatial (Y) direction giving an effective pixel scale of 0.1454 arcsec/pixel.
You may also consider binning in the spectral direction depending on how well-sampled your spectra need to be. For example, with the B600 grating and a 1" slit, the resolution is about fwhm=0.54 nm, which is equivalent to 12 unbinned pixels. Use the grating information to derive this information for other configurations.
If your science target cannot be detected in imaging mode in one of the GMOS filters in about 5 min of exposure time, you will need to supply coordinates for a nearby brighter target (R brighter than 19 and preferably a point source) and accurate offsets between that brighter target and your science target. The accuracy of blind offsetting is better than 0.1arcsec for offsets less than 20arcsec. For the blind offsetting to work it is essential that the same guide star can be reached for the bright object and the science target.
If your science target is fainter than about R=18, you will have to supply a finding chart at the time of Phase II submission.
If you are observing very faint objects in the red, you may want to consider if your program would benefit from using GMOS in Nod-and-Shuffle mode.
Are the Baseline Calibrations sufficient for your program? If you need accurate telluric line removal, you will need to add telluric standard stars to your program. If you need radial velocity standards, then these need to be added to your program.
The user may want to assess carefully the observing conditions needed for the imaging of the field. Requesting very good observing conditions for the imaging will lower the chance of getting the imaging done early during a semester, and therefore lower the overall chance of getting the MOS observations completed within a given semester.
The gaps between the three detectors in GMOS cause gaps in the spectral coverage, see the data examples. The gaps are roughly 37 unbinned pixels wide. The size of the gaps in wavelength space depends on the grating used, but is typically a few nanometers. If continuous spectral coverage is essential for your program, consider using two configurations of the grating with central wavelengths 3-5 nm different.
If you are requesting MOS observations in image quality 70-percentile or worse and your slitlets are longer than 3 arcsec, consider binning the CCDs by two in the spatial (Y) direction giving an effective pixel scale of 0.1454 arcsec/pixel (GMOS-N) and 0.146 arcsec/pixel (GMOS-S).
You may also consider binning in the spectral direction depending on how well-sampled your spectra need to be. For example, with the B600 grating and a 1" slit, the resolution is about fwhm=0.54 nm, which is equivalent to 12 unbinned pixels. Use the grating information to derive this information for other configurations.
If you are observing very faint objects in the red or need slits that are very densely packed, you may want to consider if your program would benefit from using GMOS in Nod-and-Shuffle mode.Are the Baseline Calibrations sufficient for your program? If you need accurate telluric line removal, you will need to add telluric standard stars to your program, or if possible you can add a few blue objects to your mask design. You will have to be sure that these stars are spread across the CCDs so that they cover the spectral range sampled by your observations. If you need radial velocity standards, these need to be added to your program.
The target acquisition for MOS observations is slightly more complicated than described in the generic target acquisition scenarios. The aim for the target acquisition for MOS observations is acquisition with a precision better than half the slit width over the full field covered by slit-lets. To achieve this it is necessary to include a minimum of two, preferably three, acquisition objects in the MOS mask design. It is recommended that these objects are point sources. The apertures cut in the masks for these objects are square and 2 arcsec x 2 arcsec. Further, the objects should span as much as possible of the field of view. This will allow centering of the acquisition objects in the apertures.
The acquisition objects should be bright enough to give a good S/N (larger than 200) in a one minute imaging exposure with GMOS. They should also be faint enough that the spectra of them do not significantly saturate the CCDs during the science exposures (see the MOS instructions). To ensure efficient acquisition, point sources with V magnitudes in the interval 16 mag to 20 mag are recommended for observations with grating R150_G5306. For observations with gratings B600_G5303 and R400_G5305, it is recommended to use point sources with V magnitudes in the interval 15 mag to 20 mag. Fainter acquisition objects may be used, but additional time should be budgeted for target acquisition.
As mentioned earlier, the GMOS-N CCDs show strong fringing longward of 800nm and the GMOS-S CCDs show severe fringing longward of 700nm. If the objects are faint and the spectral region of interest is located where there is severe fringing, we recomend two observing strategies to minimize the effects of fringing during the sky subtraction.
A good sky subtraction can be obtained by observing the objects in at least two positions along the slit. In this way, one can use one image to subtract the sky from the other image and obtain two sky subtracted spectra of the objects. This is a primitive way to do Nod&Shuffle.
The other method to directly do the sky subtraction and the fringe correction is to use the Nod&Shuffle method (Nod-and-Shuffle page).
In addition to the strategies listed for longslit and MOS, the user may want to consider the following for Nod-and-Shuffle observations. See also the details on the Nod-and-Shuffle page.
The overheads for Nod-and-Shuffle are significant, but can be minimized by nodding along the slit, keeping the science target(s) in the slit(s) for both the A-position and the B-position. Small nod distances have lower overheads than large nod distances. Very small nod distances (< 2arcsec) can further reduce the overheads by employing electronic offsetting.
If the program requires nodding off to sky, ie. not having the science target(s) in the slit(s) in the B-position, nodding in the q-direction as defined in the observing tool has the lowest overheads.
MOS Nod-and-Shuffle programs which are nodding along the slit need to make sure that the slit length specified in the mask design process is compatible with the total offset distance defined in the Observing Tool Nod-and-Shuffle component. Nod-and-Shuffle MOS observations should always be defined symmetrically about (0,0) in the Observing Tool Nod-and-Shuffle component.
If you are using very short slits (3 arcsec or shorter) for Nod-and-Shuffle MOS observations, it is recommened not to bin the Y-direction of the detectors.
Consider if you can take advantage of one of the configurations for which the special Nod-and-Shuffle darks will be available. If you require a special configuration, you may define a Nod-and-Shuffle dark observation in your Phase II program and it will be taken for you on a best-effort basis.
The gaps between the three detectors in GMOS cause gaps in the spectral coverage, see the data examples. The gaps are roughly 37 unbinned pixels wide. The size of the gaps in wavelength space depends on the grating used, but is typically a few nanometers. If continuous spectral coverage is essential for your program, consider using two configurations of the grating with central wavelengths 3-5 nm different.
The width of each fiber in the IFU is only 5 unbinned pixels. Thus, is it recommended to always have the Y-direction of the detectors unbinned. The spectral resolution of the IFU is equivalent to a slit with 0.31 arcsec width (4.26 unbinned pixels), thus you may consider binning in the spectral direction if your object is very faint.
You will have to decide if your science targets are best observed in IFU 2-slit mode or IFU 1-slit mode. The 2-slit mode gives you a larger field on the sky at the expense of spectral coverage. The 1-slit mode gives you larger spectral coverage, but half the field of view on the sky compared to the 2-slit mode. In 2-slit mode you will have to use one of the color filters in order to avoid overlap between the spectra, see the IFU page for details. In 1-slit mode you may have to use a filter to avoid 2nd order contamination.
In 1-slit mode the central wavelength you specify will be interpreted as the desired wavelength at the center of the detector array. In 2-slit mode, the central wavelength you specify will be the wavelength at the location of the two pseudo-slits.
If your science target cannot be detected in imaging mode in one of the GMOS filters in about 5 min of exposure time, you will need to supply coordinates for a nearby brighter target (R brighter than 19 and preferably a point source) and accurate offsets between that brighter target and your science target. The accuracy of blind offsetting is better than 0.1 arcsec for offsets less than 20 arcsec. For the blind offsetting to work it is essential that the same guide star can be reached for the bright object and the science target.
If your science target is fainter than about R=18, you will have to supply a finding chart at the time of PhaseII submission.
Are the Baseline Calibrations sufficient for your program? If you need accurate telluric line removal, you will need to add telluric standard stars to your program. If you need radial velocity standards, these need to be added to your program. Also if you need very accurate velocity calibration and you are observing in the blue where not many skylines are available from which to bootstrap your daytime CuAr, you may need to request CuAr calibrations be taken on the sky with your science observation.
Last update: December 10, 2005; Kathy Roth
Previous version: December 9, 2005; Rodrigo Carrasco