Keck LGS AO system performance

In February 2007, the Keck II AO system was upgraded with a new wavefront sensor and a new wavefront controller.

KAON 489: Performance of the Keck II AO System (2007).

A full description of the performance of the LGS AO system before the wavefront controller upgrade is presented in a PASP paper by the Keck LGS AO team.

DISCLAIMER: The results quoted here are for below average (1'') to excellent (0.3'') seeing conditions. The LGS AO system has shown to work for seeing as large as 2'', but with a large degradation in performance, as can be seen from the images of a low-elevation target (courtesy J. Lu).
1.5-2.0'' seeing (left) and 0.4'' seeing (right).

 Demonstrated observing modes
  • Tip-tilt stars to R=19.2
  • Tip-tilt stars to 72" off optical axis (OSIRIS).
  • Dither of up to 5" with LGS on-target.
  • Dither of up to 36" with LGS on-axis.
  • Position angle mode for stars R<17.5, when pupil rotation <2°/min.

 Acquisition time

This is a plot of the acquisition time for off-axis stars. The acquisition time includes all telescope and AO overheads, including slews. The data was collated by Mike Liu during the course of his brown dwarf observing program. On-axis stars can be acquired more quickly, while very faints guide stars take longer to acquire.

 Typical performance

With the implementation of the new wavefront controller, the bright guide star LGS AO performance was characterized over many nights between March and June, 2007. The Strehl ratio increased from a typical value of 0.35 to 0.39 on nights with good conditions.
Strehl vs r0 when guiding on bright tip-tilt stars.

All the results that follow were obtained with the old wavefront controller. Hence, the relative Strehl ratios are about 10% higher (e.g., 35% Strehl is now 38.5%). The faint guide star Strehls assume a dark sky.

K' filter, median seeing, on-axis
R mag. Strehl FWHM
10.0 0.35 0.052"
16.0 0.29 0.056"
17.0 0.24 0.062"
18.0 0.18 0.066"
19.0 0.08 0.120"

Strehl vs. R mag.

FWHM vs. R mag.

All the Strehl and FWHM results as a function of magnitude, seeing and wavelength are tabulated here. This includes all the data at J-, H- and K-bands taken in 2005.

Mike Liu has collected Strehl data, displayed below, in his brown dwarf program using off-axis guide stars.

 Performance as a function of zenith angle

Strehl vs. zenith angle with a bright tip-tilt guide star.
There are many factors that degrade the performance at low elevations. The distance to the sodium layer increases, so the return from the laser is reduced. The increased seeing, both on the upward and downward trajectory, results in a larger LGS spot size. The science target is imaged through more turbulence, resulting in increased atmospheric fitting and temporal bandwidth errors. The focal anisoplanatic error (cone effect) also increases. Finally, a larger spot on the tip-tilt sensor reduces the sensitivity of the sensor. Other metrics of AO performance, such as the isokinetic and isoplanatic error, also degrade with decreasing elevation.

 Isoplanatic angle

Strehl vs. isoplanatic angle with a bright tip-tilt guide star at a high elevation.
The isoplanatic angle has only been measured once on a night with excellent seeing to be 45''. The solid line in the figure represents the theoretical performance if the Strehl degradation is given by exp[-(θ/θ0)5/3] and θ0 is equal to 45''. The measurement was made by imaging a bright, on-axis tip-tilt star and moving the laser further and further away from the tip-tilt star.

 Isokinetic angle

The isokinetic angle represents the distance from the tip-tilt guide star over which the Strehl is reduced to 37% of its on-axis value. We made measurements of the isokinetic angle by measuring the Strehl ratio of a bright tip-tilt star and then of a star offset 20-40'' away from the tip-tilt star with the LGS on the offset star. The average isokinetic angle at zenith on three different nights was found to be 64'', 73'' and 95''.

 Sky coverage

sky coverage vs. galactic latitude
The fractional sky coverage at K' Strehls of 0.1, 0.2, and 0.3 is shown in the plot to the left, calculated using the galactic model of Bachall and Soneira (1980) transformed to R by Simons (1995), the on-axis K' Strehl versus R magnitude tabulated above with a 10% increase in Strehl due to the improved performance of the NGWFC, and a K' isokinetic angle of 72'' (measured on 7/26/04).

  Extended tip-tilt reference sources

Due to the nature of the tip-tilt sensor, there is a large performance penalty to be paid when using an extended source as a tip-tilt reference. In some cases, it is not even possible to close the tip-tilt loop at all. HST WFPC2 or ACAM images of several extended objects which we have attempted to guide on are shown below, with their radial surface brightness profiles displayed on the left. It appears from these experiments that extended objects must have a radial surface brightness gradient of greater >1.0 (mag/arcsec^2)/arcsec in the central 1.0" (after convolution by the seeing) in order to be used as a tip-tilt reference.

P/Temple 1, M31, and M32 radial surface brightness profiles. M31 and M32 were measured from deconvolved WFPC2 images by Lauer et al. (1998), and convolved by 0.6". The profile of P/Temple 1 was measured from ACAM images and has been offset for clarity.
Successful Unsuccessful
M32 core M31 core
P/Temple 1 (5'' FOV) Arp 220 nucleus
M82 core

The figure below shows the image quality obtained with bright binary stars with equal magnitude but different separations. There does not appear to be any significant degradation in tip-tilt performance when closing the tip-tilt loop on optical binary stars. However, there is a degradation in the performance of the image sharpening algorithm when the spacing between the stars is large, as can be seen from the PSF of the 1.95'' separation binary. The star on the left of the 1.03'' binary appears to in fact, be itself a very close binary with a separation of about 0.02''.

Sep=1.95'' (FITS)

Sep=1.22'' (FITS)

Sep=1.03'' (FITS)

Sep=0.61'' (FITS)

  Tip-tilt sensing near the moon

LGS AO observations in the vicinity of the moon lead to significant background on the STRAP tip/tilt wavefront sensor and hence a reduction in tip/tilt correction. For observations at a distance >25 degrees, the impact of the moon is negligible, while for distances <25 degrees there is a decrease in limiting magnitude of up to 2 magnitudes. If there is cirrus, the impact of the moon is much greater. KAON 425 reports all the results and gives astronomers guidelines in order to assess the impact of the moon on their observations.

  LGS AO observations near Jupiter

The fluctuating background of Jupiter light on the wavefront sensor resulted in varying low-order aberrations, including focus. KAON 385: v2 reports the results, calculations and experience on the sky trying to image Jupiter.

  Image stability

An important metric for spectroscopy is the image stability over large periods of time. The LGS AO system has differential atmospheric refraction (DAR) compensation that works by moving the position of the tip-tilt stage (TSS) as the elevation of the tip-tilt star changes. To measure the performance of DAR compensation and image stability, 60 consecutive 2s exposures were captured over a time period of 12 minutes, during which the elevation of the star increased from 39.66 degrees to 42.60 degrees.
The plots show a scatter of 7.5 mas and 2.6 mas in the X (white) and Y (red) directions respectively. However, there is no apparent drift of the star on the detector, indicating that if the color of the tip-tilt star is well-known, then the residual DAR error is negligible.

The image positioning is less accurate. The star was offset 2.5'' in the North and East directions and offset back to the starting place. The position of the star shifted by 23 mas and 7 mas in X and Y respectively. More data is being analyzed.

  Sample on-axis PSFs

The stars below were imaged on 1 Sep. 2004 (top row), and 29 July 2005 (bottom 3 rows), using the NIRC2 narrow camera (0.00994" per pixel) and the indicated filters. The LGS magnitude was V~11.0 and V~12.0, and seeing was 0.32" and 0.40" at K (r0=23 and 19 cm), respectively. The images below are the average of several 20s or 60s integrations, for a total integration time of 60s to 180s, which have been sky subtracted and divided by a flat. The displayed field of view is 1" across. The fine structure in the PSF of targets with faint tip-tilt stars (R>18) is dominated by image sharpening errors, and will vary temporally and with pupil angle. You may download a FITS file by clicking below the image.

Filter: K'

R=10.0 (FITS)

R=14.1 (FITS)

R=17.0 (FITS)

R=18.2 (FITS)

R=19.2 (FITS)

Filter: J H K'







  Sample off-axis PSFs

The stars below were imaged on UT 30 June 2005 in very good seeing conditions during the isoplanatic measurement reported above. The filter was K' on NIRC2's narrow camera. The field of view displayed is 0.6'' across.


15 arcsec off-axis

23 arcsec off-axis

  Centroid Gain for Extended Objects

Please see these notes for a discusion of how centroid gain can affect performance for extended objects.

For more information, please contact .