You are here


, Updated


GSAOI is designed to be used with GeMS at Gemini South. GSAOI is located at the f/32 output focus of Canopus and can deliver near diffraction limited images in the wavelength interval of 0.9 - 2.5 μm over a 85" x 85" field-of-view, with a scale of 0.02" per pixel. There are four gaps between the arrays of ~2.4mm, corresponding to ~2.5" on sky. It is equipped with a set of broad- and narrow-band filters, a pupil imager to set the instrument alignment and a defocus lenses to null non-common path wavefront error between Canopus and GSAOI. The GSAOI detector is composed of four Rockwell HAWAII-2RG arrays that form a 4080 x 4080 pixel focal plane. Each of the four detectors has a programmable On-Detector Guide Window (ODGW) that acts as a natural guide star (NGS) tip-tilt wave front sensor (CWFS) for Canopus. The properties of GSAOI near-infrared camera are summarized in the following sections.


Guiding Options


GeMS requires Natural Guide Stars (NGSs) to compensate for tip-tilt and plate-scale modes variation. The NGSs can be sensed using the CANOPUS Wave Front Sensors (CWFS). Ideally, three of these guide stars should be available to compensate for the plate-scale dynamical errors, however GeMS may work with 2 or even 1 CWFS with reduced performance. The observation with one or two CWFSs implies a slow degradation of the full-width half maximum (FWHM) over the GSAOI field of view, proportional to the distance to the CWFS (see Schirmer et al. 2015 for details about the GeMS performance with one NGS). The degradation in the performance is smaller than the typical anisoplanatic effect observed with a single AO system.

In July 2019, the Natural Guide Star Wave Front Sensor (NGS WFS) unit on Canopus was replaced by a new guiding unit called Natural Guide Star Next Generation Sensor (NGSNGS or NGS2). NGS2 is based on a single Electron-Multiplied CCD (EMCCD) focal plane array and uses configurable guiding windows (multi-region of interest) to read at rates up to 800 Hz.The on-sky characterization of the new guiding unit have been taken in place in October 2019. The NGS2 on-sky commissioning results show an improvement in sensitivity of about ~2.5 magnitudes, compared to the previous NGS guiding unit. The commissioning results show also a vastly improvement in acquisition and offsetting. The NGS2 Canopus WFS (CWFS) limiting magnitudes were partially verified during the on-sky commissioning in October 2019. The new limiting magnitudes for the Canopus WFS are listed in the Canopus WFS limiting magnitudes section below. More information about the NGS2 can be found in the CANOPUS Wave Front Sensors (CWFS) web page.

The focus sensing to compensate for sodium layer altitude variations with the former NGS system was obtained using CWFS3. With NGS2 this is not possible. The slow focus information is obtained using an off-axis Peripheral WFS 1 (PWFS1) star. Since the PWFS1 is not used for guiding and only to compensate the sodium layer variations, stars with fainter magnitudes can be used. The off-axis PWFS1 limiting magnitudes are given in the PWFS1 limiting magnitudes section.

Canopus WFS limiting magnitudes

The limiting magnitudes (bright and faint) of the Canopus WFSs (CWFSs) for different image quality constraints are presented in the table below. The bright limits are based on previous measurements and have not been characterized yet. The faint limits values are based on a measurements obtained during the second night of the NGS2 commissioning. The characterization of the NGS2 limiting magnitudes (faint and bright) for different sky brightness and image quality constraints will continue during the December 2019 GeMS run. Therefore, the values may be subject to change.

CWFS Limiting magnitudes (R Vega mag)
IQ CC SB Bright Faint
20%-ile 50%-ile Any 10.0 17.7
70%-ile 50%-ile Any 9.5 17.3
85%-ile 50%-ile Any 9.0 16.9

Important Note: The difference in brightness between the three CWFSs should not be larger than 3 mag. This limitation is pending confirmation and may be subject to change.

PWFS1 limiting magnitudes

The limiting magnitudes of the PWFS1 for different image quality constraints are presented in the table below. Only the faint limits are listed. The values listed in the table are an estimation and have to be verified during the next on-sky commissioning run planned for December 2019. Therefore, the values may be subject to change.

PWFS1 Limiting Magnitudes (R Vega mag)
IQ CC SB Faint Limits
20%-ile 50%-ile Any 16.70
70%-ile 50%-ile Any 16.45
85%-ile 50%-ile Any 16.20

Available guiding configurations and guide star selection

The NGSs can be sensed using only the CANOPUS Wave Front Sensors (CWFS).

The following configurations can be accepted:

  • 3 CWFS stars
  • 2 CWFS stars
  • 1 CWFS star

To compensate for tip-tilt and plate-scale modes, the NGSs should be positioned as close as possible to an equilateral triangle about the science target. Best constellations (or best asterism) are the ones that cover most of the field, or the more distant the stars are, the lower the plate-scale error will be. Care must be exercised in selecting these stars so that they remain accessible during dithered observations. Detailed information about how the constellation geometry and the guide star magnitudes can affect performance, and a description of the asterism search algorithm (MASCOT), can be found here.

When the asterism is not the optimal or the number of NGS is less than 3 NGS (2 or 1 NGS), the users might expect variations in the delivered FHWM and in the Strehl ratio across the GSAOI FoV when compared to an optimal asterism selection (see Schirmer et al. 2015 for details). Note that the performance of the system not only depend on the selected NGS constellation, the number of NGSs and their brightnesses, but also depend on the laser guide star (LGS) photons return (this parameter varies seasonally), turbulence profiling (Cn2(h)), non-common path aberrations and other AO optimization and calibration parameters. The system performance can be evaluated and visualized using the Observing Tool. Instructions of how to visualize and evaluate the performance of a selected asterism are described in the GSAOI OT Details web page.


Example of fields using NGS2 and different guide star configurations are shown below.

Example 1: 3 CWFSs

An example of an optimal natural guide stars configuration is given below. The three CWFS stars are positioned as close as possible to an equilateral triangle. The magnitudes of the CWFS range from 12.96 to 15.71 in R band. The average Strehl ratio in Ks-band, the associated error, the minimum and maximum values, calculated using the MASCOT algorithm, and the average FWHM expected for IQ=70%/CC=50%/SB=Any conditions are shown in position editor at the bottom. The Strehl map is visualized in the position editor as narrow curved green/yellow lines (see the GSAOI OT Details page for details). Note that the Strehl map displayed in the position editor does not take into account the Laser Guide Stars, however the map provides a good representation of the expected performance. The Strehl ratio and associated errors, the min, max Strehl values and the FWHM provided at the bottom of the OT Position Editor are scaled to match the value given in the GeMS performance web page. In this example, we can expect a good performance, i.e. an uniform PSF and good correction across the entire GSAOI field of view.

Example 2: 2 CWFSs

An example of 2 NGS stars configuration. Note that the use of a galaxy as a CWFS is NOT an option. The magnitudes of the two available CWFSs are 10.6 and 11.6 in R band. The expected average Strehl ratio and FWHM in Ks-band for IQ=70%/CC=50%/SB=Any are 11.3% and 0.09 arcsec, respectively. Note that because the Strehl map does not take account the location and contribution of the Laser Guide Stars, the lines shown in the position editor do not reflect the true Strehl variation across the GSAOI field of view. However, the values displayed at the bottom of the Position Editor are correct. The correction will be better close to the two CWFS stars and degraded toward the edge of the GSAOI field view. Therefore you may expect a lower performance across the GSAOI field of view compared to an asterism with 3 CWFS: a fairly uniform PSF but with increasing FWHM values, proportional to the distance to the CWFSs.

Example 3: 1 CWFS:

An example of one NGS star configuration. Left: DSS image of the Gemini Frontier Field galaxy cluster MACSJ 0416.1-2403 displayed in the OT position editor. The magnitude of the CWFS is 13.4 mag in R-band. The expected average Strehl and FWHM in Ks-band for IQ=70%/CC=50%/SB=Any are 6.9% and 0.090 arcsec, respectively. The correction will be better close to the WFS star and degraded toward the edge of the GSAOI field view. Therefore you can expect a fairly uniform PSF, but with variable and poorer FWHM across the GSAOI field of view toward the edges. Note that only point-like sources can be used as NGS. Right: Ks-band image of the galaxy cluster MACSJ 0416.1-2403 observed with GeMS/GSAOI (Schirmer et al. (2015, Fig. 8). As expected, the PSF is fairly uniform, but the FWHM and Strehl ratio show variations toward the edges of the image. The FWHM increases from 72 to 122 mas while the Strehl decreases from 15% to 3% from the location of the NGS. The average Strehl ratio and FWHM shown in the image are consistent with the values provided by the MASCOT algorithm in the OT.

 Fig.8 from Schirmer et al. (2015)

On detector guiding


Each of the GSAOI detectors contains programmable ODGWs. The ODGWs provide tip-tilt information for up to four natural guidestars to GeMS. The ODWGs can be read out at up to 800 Hz during a science exposure. Sizes of either 4x4, 6x6, 8x8, 12x12, 16x16, 32x32, or 64x64 pixels can be specified, although most science observations will be completed using 6x6 or 8x8. The ODGWs are used to measure star centroids for either Canopus tip-tilt correction or for fast guiding to correct differential flexure between GeMS and GSAOI.The figure below shows a typical ODGW dark (left) and flat-field (right). As one can see in the figure, only immediately adjacent pixels in the science image are affected.

On-Detector Guide Window with the main array exposed for a dark frame (left) and a flat-field frame(right).

Each of the four ODGWs (one per detector) can be read out with different exposure times, as long as each exposure time is an integer multiple of the shortest exposure time. In practice, the four ODGWs are read out synchronously using non-destructive reads. These data are processed in a way to synthesize exposure times that are an integer multiple of the basic read time.

The ODGWs are read out using Fowler Sampling. The minimum frame time, T, to read an NxN pixel ODGW with M-M Fowler non-destructive reads (NDRs) is given by the following relation:

T = (2M + 1)[5.0N2 + 37.06 N + 17.37] μs

here, 2M + 1 is number of passes through the whole array; 1 reset pass, M Fowler NDRs at the start of the integration, and M Fowler NDRs at the end of the integration. Each pass through the array takes 5.0N2 + 37.06N + 17.37 μs, which consists of 5.0 μs/pixel to read the NxN ODGW pixels, a row overhead of 37.06 μs/row, and a frame of overhead of 17.37 μs. A set of simulations were performed to determined the exposure times needed to operate the ODGWs at fixed frame rates (likely to be used in practice). The derived parameters have been used to simulate the ODGW centroiding accuracy for stars of different brightnesses and for different filters. These values will be used as starting point for the ODGW characterizations during GSAOI + GeMS commissioning.

Reads of the ODGWs are interleaved with reads of the science image during resets and Fowler sampling. This requires a reduction of the ODGWs frame rate to ~200 Hz during these periods when the science image is accessed. Otherwise, the science image read out time would be unacceptably long.

On-Detector Guide Window Sensitivities

The On-Detector Guide Window performance has been estimated using the image performance model (see GSAOI Sensitivity section) to generate star frames that have then been centroided and the RMS centroiding accuracy determined from 200 simulated guide star frames. Limiting magnitudes that achieve a centroiding accuracy of ~ 2 mas (i.e., 0.1 pix) in integration times of 0.01 sec. (i.e., when used as a GeMS NGS) and 30 s (i.e., when monitoring flexure variations) are listed in table below. The table also list the magnitude of the detector saturation for both cases. Note that the values listed in the table are a good approximation of what it is expected, but will be confirmed during GSAOI commissioning.

 On-Detector Guide Window Sensitivities (2 miliarcsec RMS)

Lim. magnitudes (mag)

10 ms integration

Sat. magnitudes (mag)

10 ms integration

Lim. magnitudes (mag)

30 sec integration

Sat. magnitudes (mag)

30 sec integration 

Z 15.1 8.0 22.7 16.7
J 14.2 7.1 21.2 15.8
H 14.5 7.3 21.0 16.0
K(prime) 13.7 6.6 20.4 15.2
K(short) 13.6 6.5 20.3 15.2
K 13.5 6.5 20.2 15.1
J-continuum 12.0 4.6 20.0 13.5
H-continuum 12.0 4.7 19.7 13.4
CH4(short) 13.5 6.3 20.5 15.0
CH4(long) 13.1 5.9 20.1 14.6
  K(short) continuum 11.2 4.1 19.1 12.7
  K(long) continuum 11.0 3.8 18.8 12.5
He I 1.083 μm 12.3 5.3 20.6 13.8
H I Pγ 11.9 4.6 20.2 13.2
H I Pβ 11.8 4.6 19.4 13.3
 [Fe II] 1.644 μm 11.8 4.5 19.4 13.2
H2O 12.3 5.1 19.8 13.8
He I (2p2s) 11.2 3.9 18.9 12.7
H2 1-0 S(1) 11.3 4.1 19.0 12.8
H I Brγ 11.3 4.0 18.9 12.7
H2 2-1 S(1) 11.0 3.8 18.9 12.6
CO Δv=2 11.6 4.4 18.8 13.2

On-Detector Guide Window Availability

The availability of suitably faint guide star in the near-infrared was investigated by Spagna, A. (2001, STScI_NGST-R-0013B, “Guide Star Requirements for NGST: Deep NIR Star Counts and Guide Star Catalogs”) for the NGST (now JWST). Spagna tabulated cumulative star counts from which probabilities can be calculated that the 2' diameter GeMS field for any science object will contain at least one guide star brighter than a particular limit. The results of this investigation are presented graphically in the figure below. The figure shows that the probability of finding at least one K ≤ 13 mag guide star, suitable for use as a GeMS NGS with the broad-band filters, in the GeMS field is > 11% while the probability of finding at least one K ≤ 20 mag guide star, suitable for flexure monitoring with the broad-band filters, in the GeMS field is high (> 97%).


Guide star acquisition probabilities at J (solid lines) and K (dashed lines) for Galactic latitudes of 30o (triangles), 60o (diamonds), and 90o (circles).


This section presents the expected imaging sensitivities for GSAOI. The first table lists the limiting magnitude, the saturation magnitude, the assumed Strehl ratio and the sky brightness for each filter. The second table lists the background contribution from different sources for all filters. The imaging sensitivities presented here are theoretical and have been obtained using a model for the object (point-source) and background signals. These estimated values are a good approximation of what it is expected for GSAOI. Accurate sensitivity values will be derived during the GSAOI and GeMS commissioning. For the modeling, the following parameters were assumed:

  • Detector dark current of 0.05 e-/s/pix, a read noise of 5 e-, using multiple reads and detector quantum efficiency of ~ 80%
  • Background signal considerations: airglow emission and thermal emission from the sky, telescope, cryostat window, cryostat interior and GeMS.
GSAOI Image Sensitivities: 1 hour on-source integration, 0.4" natural seeing through 0.08" x 0.08" square aperture and photometric conditions (50%-il Cloud Cover)
Filter Limiting Magnitude (mag) Saturation Magnitude (mag)(1) Assumed Strehl Ratio Sky Brightness (mag/arcsec2)
Z 25.6 14.8 0.2 17.1
J 24.1 13.9 0.2 14.9
H 24.1 14.0 0.4 14.0
K(prime) 23.9 13.3 0.6 13.5
K(short) 23.8 13.2 0.6 13.4
K 23.7 13.2 0.6 13.3
J-continuum 23.1 11.7 0.2 15.0
H-continuum 22.9 11.5 0.4 14.1
CH4(short) 23.6 13.1 0.4 13.9
CH4(long) 23.4 12.7 0.4 13.8
K(short) continuum 22.7 10.9 0.6 13.6
K(long) continuum 22.5 10.6 0.6 13.5
He I 1.083 μm 23.7 12.0 0.2 16.1
H I Pγ 23.5 11.5 0.2 16.2
H I Pβ 22.5 11.5 0.2 14.0
[Fe II] 1.644 μm 22.7 11.4 0.4 13.8
H2O 23.3 11.8 0.6 13.9
He I (2p2s) 22.5 10.8 0.6 13.3
H2 1-0 S(1) 22.6 10.8 0.6 13.4
H I Brγ 22.6 10.7 0.6 13.5
H2 2-1 S(1) 22.5 10.6 0.6 13.5
CO Δv=2 22.4 11.2 0.6 12.6

(1) Saturation magnitudes are estimated assuming a minimum integration time of 5 sec and detector full well depth of 50,000e-

GSAOI Image Background Contribution (600 sec integration)(*)

Airglow (e-/pix)

Sky Thermal (e-/pix) Telelescope Thermal (e-/pix) GeMS Thermal (e-/pix) Window Thermal (e-/pix)  Total  (e-/pix)
 Z  3867 0 0 0 0 3897
 J  17904 0 0 0 0 17934
 H  39048 1 5 16 1 39101
 K(prime)  22167 378 2162 6652 428 31819
 K(short)  21285 447 2555 7860 507 32685
 K  19069 1002 4249 13072 849 38273
 J-continuum  1807 0 0 0 0 1838
H-continuum  3216 0 0 0 0 3246
 CH4(short)  15301 0 0 1 0 15332
 CH4(long)  14851 0 2 7 0 14891
 K(short) continuum  2468 13 106 327 21 12966
  K(long) continuum 430 107 579 1782 116 3045
 He I 1.083 μm 796 0 0 0 0 826
H I Pγ 462 0 0 0 0 492
 H I Pβ 4611 0 0 0 0 4641
 [Fe II] 1.644 μm 4045 0 0 1 0 4076
H2O 4730 87 106 327 20 5302
 He I (2p2s) 3183 45 74 230 15 3577
 H2 1-0 S(1) 3024 7 146 449 29 3685
 H I Brγ 2285 15 221 680 44 3275
H2 2-1 S(1) 946 63 484 1491 97 3112
 CO Δv=2 251 1228 2718 8360 549 13134

(*)  Model background signals for each filter assume a dark current of 30 e-/s/pix in 600 sec integration.

Note: the model background signals for each filter presented in the table have to be verified and confirmed during the GeMS + GSAOI commissioning.