Ground layer wavefront reconstruction using dynamically refocused Rayleigh laser beacons

1
Ground layer wavefront reconstruction using
dynamically refocused Rayleigh laser beacons

• C. Baranec, M. Lloyd-Hart, M. Milton, T. Stalcup,
  M. Snyder, N. Putnam and R. Angel
• Center for Astronomical Adaptive Optics
• Steward Observatory, The University of Arizona

OSA 2005 Adaptive Optics Analysis and Methods
2
GLAO - Introduction

• Ground layer adaptive optics (GLAO) correction is
  a method for correcting the wavefront errors
  caused by turbulence close to the telescope.
• By using a constellation of guide sources, one
  can average the measured wavefronts, giving an
  estimate of the ground layer turbulence.
• Applying this correction to a DM conjugated near
  the ground, removes the wavefront aberration
  common to a wide field.
• With varying measurements of the ground layer
  turbulence being up to 2/3 of the total
  turbulence, this can greatly improve seeing over
  this same field.

3
GLAO at the MMT

• GLAO will be beneficial for current and future
  extremely large telescopes (ELTs). It promises
  partial wavefront correction and uniform PSFs
  over a wide field of view.
• GLAO is a powerful new technique that needs
  experimental validation.
• We are investigating GLAO as we move forward to
  testing new AO techniques for ELTs at the MMT.
• We have deployed a five beacon Rayleigh laser
  guide star (RLGS) source at the MMT to test
  ground layer and tomographic reconstruction of
  atmospheric turbulence. Here, I present our
  system and first results in relation to ground
  layer adaptive optics.

4
RLGS Beam Projector at the MMT

• Two 15 W doubled YAG lasers at 532 nm pulsed at 5
  kHz.
• The laser beams are combined with a polarizing
  beam splitter.
• A computer generated hologram splits the combined
  beam into 5 beams that are projected onto a
  circle of 2 arc minutes diameter.
• Projection optics mounted on the telescope axis
  behind the secondary mirror
• Photometry
• Measured 760,000 ph/m2/J
• Typical Sodium LGS 840,000 ph/m2/J (J. Ge 1998)

5
Wavefront Sensor Instrument

• Wavefront Sensor (WFS) Instrument mounts to MMT
  Cassegrain focus. Run both RLGS and NGS
  simultaneously.
• RLGS WFS
• Multiple laser guide star Shack-Hartmann
  wavefront sensor.
• Hexapolar geometry, breaks pupil into 36
  subapertures.
• Uses a range gated Lincoln Labs CCID18 chip run
  at 55 Hz.
• Dynamic refocus system removes the focus term
  from each pulse of the RLGS over its range gate
  from 20 30 km
• NGS WFS
• Optical clone of the MMT-AO NGS WFS camera with
  an E2V CCD39 run at 110 Hz.
• Pupil broken into 12x12 subapertures of which 108
  are illuminated.
• Sensor on translation slide, to allow exploration
  of field in one axis.
• RLGS/NGS WFS Synchronization
• externally controlled LED flashers used to
  synchronize data capture for both RLGS and NGS
  WFS. Flashed once per second.

6
LGS WFS Data

• Shack-Harmann patterns of the five beacons on the
  RLGS WFS after background subtraction.
• Windshake of the secondary mirror hub bends the
  telescope, causes patterns to move around.
• Flashes due to LED synchronization.
• Used physically constrained iterative blind
  deconvolution methods to measure spot positions
• Data Quality.

7
Wavefront Reconstruction

• Wavefront reconstruction of the ground layer
  turbulence and the ground truth natural star
• RLGS wavefront reconstruction by inversion of
  synthetic influence matrix of Zernike modes on
  our geometry of Shack-Hartmann pattern.
• Estimate of ground layer turbulence by averaging
  the Zernike coefficients of each beacon.
• NGS wavefront reconstruction by using the same
  reconstructor matrix as used in the closed-loop
  MMT AO system. The NGS WFS is optically the same,
  so we can use the same reconstructor.
• Estimate of GLAO performance by subtracting
  ground layer estimate from NGS ground truth.

8
Performance with Field Angle

• Exploration of GLAO performance with field angle.
• Figure shows the position of the NGS for each
  data set in relation to the RLGS. Data taken over
  a period of 2 hours.

9
Reconstructed Data

• Phase reconstruction of ground layer estimate and
  NGS Zernike orders 2-6.
• Upper row Shack-Hartmann patterns from RLGS and
  NGS.
• Bottom row Reconstructed phase from ground layer
  estimate and NGS. In good agreement but show
  differences due to non-common turbulence and
  measurement error.

RLGS NGS
10
Zernike Mode Tracking

• An example comparison of three Zernike modes
  between GLAO estimate and NGS ground truth.
• NGS in dashed blue.
• GLAO average of the five RLGS in solid black.

Each sequence is approximately 3 seconds.
11
Residual RMS after correction

• Example RMS wavefront aberration over 3 seconds
  for Zernike orders 2 through 6
• NGS in blue.
• Average RLGS in black.
• Residual wavefront aberration of NGS after GLAO
  correction in red.

NGS RMS wavefront aberration 650 nm Residual NGS
RMS wavefront aberration after correction 380 nm
12
Performance with Field Angle

• RMS stellar wavefront aberration in nm, averaged
  over the modes of each Zernike order. Before
  correction, top, and after GLAO correction,
  bottom. Median seeing at the MMT at 500nm is ro
  15cm, so we were working under poor seeing
  conditions.

Zernike order Set 1 Set 2 Set 3 Set 4 Set 5
2 462 572 513 571 559
2 (after correction) 255 316 308 349 343
3 308 404 365 383 379
3 (after correction) 198 283 226 246 258
4 223 285 261 276 269
4 (after correction) 142 181 168 184 190
5 183 220 207 220 220
5 (after correction) 140 166 152 168 168
6 159 184 175 194 170
6 (after correction) 116 143 130 154 143
2-6 645 809 732 797 778
2-6 (after correction) 397 487 463 518 518
ro (cm) _at_ 500 nm 12.1 9.0 10.3 9.2 9.8
13
Performance with Field Angle

• GLAO performance as a function of field angle

Over the course of taking data, ro varied from
9.0 to 12.1 cm at 500 nm. To allow direct
comparison, all data points have been rescaled to
the MMTs median seeing of ro 15cm at
500nm. Bars on left show the uncorrected
measured NGS RMS wavefront error rescaled to ro
15cm.
14
Performance with Field Angle
15
Ground/Upper layer turbulence
From Hardy (1998), the power in Zernike orders
2-6 is given by The overall ground layer
corrected residual wavefront error inside the
beacon constellation is 356nm. This yields values
of r0 for the ground and upper layers Uncorrected
upper layers r0 30 cm Ground layer r0 19
cm An approximate division of 2/3 power in the
ground layer, and 1/3 power in the free
atmosphere. In agreement with other studies done
at Cerro Pachon.
16
Ground Layer Isoplanatic Angle

• From our data we were able to calculate other
  atmospheric parameters.
• For each of the five data sets, we were able to
  find the residual RMS stellar wavefront
  aberration using each individual beacon as a
  correction.

• This gave us 25 measurements of RMS residual
  error as a function of angle.
• Plotting these points and fitting a curve of the
  form
• y a b ?0-ground 5/3 gave us a measurement of
  ?0.
• We found ?0-ground 29 arcsec at 500nm.

Beacon NGS Separation (arc sec)
17
Mean Height of Ground Layer

• Given our measurements of ?0-ground and
  r0-ground, we can calculate the mean height of
  the ground layer turbulence, h. From Hardy (1998)

With a mean sec(? ) 1.05 for these
observations, we calculate h 445 m
18
Conclusion

• What we have learned about GLAO correction
• Using five Rayleigh laser guide beacons, we can
  get a measurement of the ground layer turbulence.
• The residual RMS stellar wavefront aberration
  after correction is more constant in time.
• Ground layer correction is relatively flat within
  the diameter of the RLGS constellation with a
  gradual decay of correction outside.
• Gives modest seeing improvement even into I band.
• Most importantly We have seen an average 40
  improvement in wavefront error over a 2 arcminute
  field.

19
Future Work

• Another run at the MMT next week with much
  improved instrument
• New CCD for the RLGS that actually works
  properly!!!
• Made a number of optical improvements to system,
  easing alignment and increasing throughput.
• Upgraded the RLGS WFS from 36 to 60 subapertures,
  allowing wavefront reconstruction up to Zernike
  order 9.
• Will allow us better understanding of GLAO
• Will allow us to take the next step and attempt
  tomographic reconstruction of the atmospheric
  turbulence
• Future work
• With data collected next week, we will be
  preparing to run the system in closed loop with
  the MMTs adaptive secondary later this year
• See Michael Lloyd-Harts talk on Development of
  Multi-Laser Guide Star Adaptive Optics Techniques
  for Extremely Large Telescopes

20
Dynamic Refocus in Action

• RLGS Shack-Hartmann patterns with and without
  dynamic refocus (DR) running.
• Without DR, off-axis spot elongation. Can be seen
  here as radial streaking of spots.
• Data taken 29th Sept 04, 1128 pm.

21
Details of Beam Projector
22
WFS Instrument Optical Layout

• (1) Wide field imaging optics and camera, (2)
  Dichroic mirror, (3) Natural guide star wavefront
  sensor optics, (4) Closeup of NGS WFS camera, (5)
  Dynamic refocus resonator and optics, (6)
  Rayleigh Laser guide star wavefront sensor arm,
  (7) Closeup of RLGS WFS Camera.

23
Estimated FWHM

• Given our GLAO correction of Zernike orders 2
  through 6, and assuming perfect tip/tilt
  correction, we can calculate the FWHM of a long
  exposure image using our current system.
• For an ro 15cm at 500nm, we can see the
  comparison of the seeing FWHM and the FWHM after
  correction for bands in the near IR.
• H and K bands are nearly diffraction limited, and
  there are significant gains in FWHM into I band.

Band Wavelength r0 Seeing Diffraction FWHM after correction
µm m arcsec arcsec arcsec
K 2.2 0.824 0.551 0.0698 0.0733
H 1.6 0.562 0.587 0.0508 0.0592
J 1.25 0.418 0.617 0.0397 0.116
R 0.9 0.282 0.658 0.0286 0.229
I 0.7 0.209 0.691 0.0222 0.268
24
Substandard Data

• Data quality from previous runs was substandard.
  Due to a number of factors
• Our RLGS WFS CCD was horrible
• Bad MTF caused images on RLGS WFS to look
  terrible.
• Typical FWHM of Shack-Hartmann spots found to be
  3.7 arcsec. When measured on separate camera was
  1.5 arcsec.
• Lots of Noise / Fixed pattern Noise
• Video Dropouts
• Vastly different amplifier biases
• Found our alignment tolerances were very tight
  and made it difficult to align in short amount of
  time we had on the mountain.
• Typical problems getting a prototype system up
  and running
• Working in 40 mph winds, which made us stop
  observing early.