Change page style: 

Observing Strategies

Given the complexity of the AO system, careful designing of GeMS/GSAOI observations is essential to achieve the desired performance required for the success of a science program. In this page we provide important information to guide the users with defining the observing strategies to follow with GeMS/GSAOI. The page also provides some guidelines to maximize observing efficiency and avoid some common errors. This information might be also useful to prepare a GeMS/GSAOI proposal submission (Phase I).

The users may want to take into account the following factors when designed observing sequences:




Natural Guide stars (NGS)

As it is mentioned in Guiding Options page, the best and most uniform AO correction is achieved when the NGSs are positioned as close as possible to an equilateral triangle. The best asterisms are the ones that cover the larger portion of the field, and the more distant the stars are from each other, the lower the plate scale error will be. When the asterism is not the optimal or less than 3 NGS (2 or 1 NGS) are used, the users might expect larger variation in the delivered FHWM across the GSAOI FoV when compared to an optimal asterism selection, affecting the PSF uniformity across the field. If using only one NGS, the user can expect corrections slightly better compared to a single conjugate adaptive optics system (e.g. Altair). This is because the distribution of the 5 laser guide stars (LGS) on the sky (a 1' x 1' square) can better compensate the variation across the GSAOI FoV compared to a classical AO system with a single laser.

  • The users should pay special attention to the guide star selection. Sometimes the available optical catalogs show the wrong R magnitude. If you have doubts about the magnitudes of the stars, always use the brighter stars available inside the Canopus patrol field. Rotating and shifting the GSAOI FoV may help with selecting a better asterism.
  • Only the Canopus NGSs (CWFS) can be used for fast tip/tilt correction. The GSAOI On-Detector Guide Windows (ODGW) are not yet implemented, and therefore not allowed for fast tip/tilt correction.
  • Only one ODGW star can be selected for flexure correction.
  • Always assign the brightest guide star inside the Canopus patrol field to CWFS3.



Recommended maximum exposure time for individual frames

For broad band filters (Z J H K Kp Ks), the maximum recommended exposure time for a single frame is 120 sec. Longer exposures are not recommended because the sky changes significantly between exposures, in particular in H, Kp, Ks and K bands, making proper sky subtraction difficult.

For narrow band filters, the maximum recommended exposure time is about 360 seconds. Make sure the exposures are long enough to be background limited. The users have to use the Faint Object or Very Faint Object read modes to observe with narrow band filters (see below).

Science programs that need separate sky frames should alternate the science observations with the sky frame observations. It is strongly recommended to observe a set of sky frames every 10 minutes (less is better), in particular for H and K filters (see below for details about sky frame observations).




Read mode selection, co-adds and associated overheads

The selection of the read mode and the number of co-adds used in any GeMS/GSAOI observation will directly impact in the overheads. It is important to balance between the read mode and the co-adds needed for your observation to minimize the associated overheads and optimize the S/N ratio of the resulting image. The overheads associated to each read mode are listed in this table. Note that in the Observing Tool GSAOI component, the read mode is automatically selected based on the filter and exposure time used. The user can select a different  read mode using the OT GSAOI sequence iterator.

  • If the observations are photon dominated, the Bright objects read mode may be used. If the objects in the field are too bright, then you can reduce the exposure time and add some co-adds to avoid saturation. However, the penalty will be the increase of the overheads due to the readout time associated to each co-add (see this table for details).
  • The minimum exposure time with GSAOI using the Bright Object mode is 5.3sec.
  • If the observations are background-limited, the Faint or Very faint object read modes may be used. The choice between the two read modes depends on the filters used and the exposure time of each frame. Note that the recommended exposure time per filters is automatically displayed in the GSAOI component. For broad band filters, it is recommended to use Faint Object read mode. For narrow band filters, it is recommended to use the Faint or Very Faint Object modes, depending on the exposure time. The readout noise can be reduced by using different read modes. For example, you can be tempted to observe with broad band filters and using the Very Faint Objects read mode to reduce as much as possible the read noise. The difference in read noise between Faint and Very Faint modes is ~3 e-. However, the penalty is an increase in the read out time by a factor of two, from 26.2 sec to 47.7 sec. Therefore, short exposure times (< 30 sec) with Very Faint Object read mode is not a very efficient combination. 




Dithering and offset size

The GSAOI FOV is formed by four HAWAII-2RG 2k x 2k arrays mounted in a 2 x 2 mosaic. Note that the arrays are not perfectly parallel to each other and the size of the gaps varies between 2.5" and 3.5" on the sky. Since most observations will require to dither in order to prevent any loss of information due to objects falling in the gaps, it is recommended that the dithering pattern is constructed with a minimum offset size of 4" between images.

When an observing sequence is constructed, the user has to keep in mind the following:

  • The CWFS NGSs must remain inside the patrol field area of Canopus in all steps defined in the GSAOI offset iterator. Note that in the case that one or more CWFS NGS move outside the Canopus patrol field area during dithering, an error message will be appear in the OT. A visual inspection of the offsets can be done using the OT Position Editor (see Visualization of Selected Asterism page for details).
  • The star used for flexure correction (ODGW) must remain inside the area defined by the array where the star was selected. In the case that the ODGW move outside the allowed area, an error message will be appear in the OT. The ODGW patrol field is  visualized in cyan (dashed lines) in the Position Editor and can be activated using the ODGW button in the GSAOI Position Editor.
  • Remember, the overheads associated to dither, i.e. the time needed to open all loops, move the telescope and close all loops again, is 30 sec per dither position.
  • Note about precise astrometry with GeMS: some programs require a good astrometric accuracy (see " GeMS: first on-sky results " by F. Rigaut et al., SPIE, 2012, Vol. 8447, pp. 84470I-84470I-15A for a detailed analysis of the astrometric performance obtained with GeMS). In this case, we don't recommend to dither. The complicated field distortion introduced by GeMS makes it difficult to obtain a good astrometric performance in a combined image when offsets are used. If the user needs to dither, then we suggest to use offsets no larger than 0.1- 0.2 arcsec (5 - 10 pixels).



Observing sky frames

As any near-IR imaging, GeMS/GSAOI observations require background subtraction. In the case of crowded fields, fields embedded in a large nebulosity or fields containing an extended object, sky frames have to be acquired using a blank region near the science target. In the case of sparse fields, the same science images can be used to create a master sky frame to subtract the background as long as enough dither positions are obtained to mask the objects in the individual frames.

A. General notes

When obtaining separate sky images, the dither pattern should be large enough to remove point sources in the combined sky frame (5"-10" is usually more than enough since the observations are unguided). Preferably, obtain the sky frames using the same exposure time and number of exposures as used for your science observations. Exposures shorter (or longer) than your science observations can make the sky subtraction (and your data reduction) more difficult. Avoid saturating objects in your sky frames when possible - remember, these frames are unguided, so feel free to move the center of the field around to exclude the brighter stars. Saturated objects are very difficult to remove and can affect the background subtraction. Observe sky frames every 10 minutes (less is better), particularly when using the H and K broad band filters.

B. Sparse fields

In the case of sparse fields, the science images can be used to construct a master sky frame to subtract the background. The offsets should be large enough to remove point sources and cover the gaps when making sky images. If the field contains sources that are more extended than point sources, it is recommend to construct a dither sequence with offsets slightly larger than the average size of these objects.

C. Crowded fields and fields with extended objects

For crowded fields, fields embedded in a large nebulosity or fields containing one or more extended object, sky frames have to be acquired in a blank region near the science target. In some cases (extended nebulosities, large star clusters) very large offsets are required to reach a blank field to observe the sky frame.  If the offset to reach the sky frame is larger than 5 arcmin from the science position, the laser has to be shuttered and the science field will need to be re-acquired when the telescope is back from the sky, incurring in additional overheads to the program. Depending on the location of the blank field, the best observing strategy to follow is different.

C.1. Sky field located < 5 arcmin from the science target

  • The sky frames can be interleaved with the science frames and included in the same observing sequence. Note that the entire sequence, including overheads, cannot be longer than 2.5 hours. A sequence longer than 2.5hrs requires an additional acquisition.
  • If the science sequence includes more than one filter, the exposures are short and the sequence through all filters is not longer than 10 minutes (including overheads), then it is recommended to first observe the science target with all filters from blue to red, then move to the blank field and observe the sky frames in all filters in the reverse order. e.g. science (J) + science (H) + science (K) + sky (K) + sky (H) + sky (J). This strategy optimize the observations and minimize the overheads.
  • If the science sequence includes more than one filter and the exposures are long, then it is recommended to observe one filter at time, e.g. science (J) + sky (J) + sky(H) + science (H) + science (K) + sky (K), within the same sequence. Note the use of the traditional ABBA pattern (science-sky-sky-science) in order to minimize offset overheads.

C.2. Sky field located > 5 arcmin from the science target

In this case, the laser propagation must be stopped due to Laser Clearing House restrictions. Therefore, when returning to science target, the LGSs have to be re-acquired and the CWFS NGS acquisition re-checked.

  • For observations with only one filter, the best strategy is to create a sequence containing one science and one sky blocks, of as longer duration as scientifically useful (as noted above, it is recommended to obtain a sky block every 10min or less of science). As many sequences as required to obtain the total time on-source should be provided. Note that it is not feasible to observe science-sky-sky-science in this case, since there can be a gap of more than 10min between the second sky block and the start of the second science block as the target (LGS+NGS) is re-acquired.
  • For sequences with short total time and multiple filter observations, the best strategy is to observe the science field in all filters and then move to the sky position. The sky observation must be defined in a separate sequence, and setup unguided.
  • For long sequences with multiple filters observations, the best strategy is to observe the science field then move to sky for each filter. In this case, one science sequence and one sky sequence have to be created separately for each filter.