Sie sind auf Seite 1von 11

AT THE WORLDS LARGEST TELESCOPES, ASTRONOMERS ARE INCREASINGLY

ABLE TO CANCEL OUT THE BLURRING CAUSED BY EARTHS ATMOSPHERE

AND VIEW THE UNIVERSE ALMOST AS SHARPLY AS IF THE TELESCOPES WERE

IN OUTER SPACE. BUT ITS NOT EASY. BY GOVERT SCHILLING

On Friday, November 26, 1999,


Scott Acton saw an erupting volcano.
You might think thats no big deal if you Acton is a member of the Keck Adaptive
live where Acton does, near the volcanoes Optics Team. Since the Keck II telescope
of Hawaii. But this one erupted on anoth- achieved AO first light on February 5,
er world, 620 million kilometers away. The 1999, the system has been used to study a
glowing eruption appeared as small as the variety of astronomical objects, demon-
period at the end of this sentence seen strating its enormous potential. On No-
from a distance of two kilometers. And vember 26th of that year Shrinivas Kulkar-
most remarkable of all, the observation ni (Caltech) was about to observe the close
was made through high cirrus clouds. companion of the dwarf star Gliese 569,
Welcome to the world of adaptive op- but cirrus clouds above Mauna Kea made
tics (AO). After years of promise and de- this impossible. Shri decided to go drink
velopment, AO is revolutionizing ground- some coffee with a colleague, says Acton,
based astronomy at many of the worlds who was in the Keck II dome to assist with
largest observatories. It is technologys the AO observations. We wondered if
triumphant answer to an age-old prob- there was anything else we could observe.
lem: the blurring of telescopic images by Jupiter was one of the few objects
atmospheric turbulence, which astronomers call the seeing. By bright enough to show through the clouds, so Acton and his
using elaborate wavefront sensors, fast computers, and rubber colleagues turned the giant telescope to the giant planet and
mirrors that reshape themselves hundreds of times a second, aimed at Io, the innermost of Jupiters large satellites. We
astronomers are able to untwinkle the stars and see almost locked the AO system on Io, he says, and it happened to have
as sharply as if the atmosphere didnt exist. this active volcano! From three near-infrared images, we con-
Since the early 1990s many 4-meter-class telescopes have structed a false-color picture of Io, showing a wealth of surface
been outfitted with AO systems. In the past three years larger details. The smallest markings were just 150 kilometers
telescopes have joined the movement, notably the two 10-meter across. For Acton, bad-night astronomy turned into a miracle.
Keck telescopes, the 8.2-meter Gemini
North reflector, and the 8.3-meter Subaru
telescope, all at Mauna Kea, Hawaii. Soon Above: Jupiters moon Io was 1.2 arcseconds wide when the Keck II telescope and adaptive-optics
AO will be available to astronomers at all system imaged it through thin clouds on November 26, 1999. A hot volcanic eruption stands out on
of the worlds major observatories. Mean- Ios limb in this three-color infrared image. Dozens of surface features can be identified here with
while, optical engineers are devising ways the help of views from the Voyager and Galileo spacecraft. Courtesy Keck Adaptive Optics Group.
to broaden AO techniques to make them Right: Making an artificial star. An adaptive-optics system needs to watch a bright guidestar in its
work in more situations and produce narrow field of view to measure atmospheric distortions. If no bright star is handy, the Calar Alto
wider fields of view than they do now. Observatory in Spain can put one there. A powerful laser on the side of the observatorys 3.5-meter
With a little luck, these efforts will ulti- telescope excites sodium atoms about 90 kilometers up. The sodium glows like a pinpoint as seen
mately open the way to a new generation by the telescope. The adaptive-optics system measures the atmospheric distortions of this pin-
of 25- to 100-meter ground-based tele- point and cancels them out, sharpening the view of the universe beyond. (On this particular night,
scopes, something once thought impossi- the laser beam was stopped by a layer of thin clouds.) Courtesy Max Planck Institute for Astronomy
ble (S&T: August 2000, page 52). and Extraterrestrial Physics.

30 October 2001 Sky & Telescope 2001 Sky Publishing Corp. All rights reserved.
2001 Sky Publishing Corp. All rights reserved.
Back into Focus ter than an amateurs backyard 10-inch.
When Galileo Galilei caught his first glimpses of the Jovian At least not until recently. Adaptive optics can take a
satellites almost four centuries ago, he could hardly have imag- smeary, seeing-confused mess of a star image and put it back
ined that a 300-ton descendant of his small telescope would together, nearly achieving diffraction-limited perfection by
resolve surface features on them. Galileos homemade tele- optically counteracting each and every atmospheric distortion
scope had a very poor lens only 4.4 centimeters (1.7 inches) in of the stars light at every moment.
diameter, and he couldnt see details smaller than about 10
arcseconds wide the angle subtended by a dime 300 meters
away. For instance, Galileo couldnt resolve Saturns rings.
If Galileos lens had been perfect, it would have resolved de-
tails 3 arcseconds wide the theoretical angular resolution of
a lens 4.4 cm wide. A telescopes resolution is inversely propor-
tional to its aperture, so a 13-cm instrument resolves details as
small as 1 arcsecond, while a 130-cm has a theoretical resolu-
tion of 0.1 arcsecond. So why is it remarkable that the 10-
meter Keck II should see markings on Io? With its aperture,
the telescope ought to resolve details as fine as 0.013 arcsecond.
If Keck were in orbit or on the Moon, it might. But as every
telescope user knows too well, the Earths atmosphere wreaks
havoc with highly magnified images. Air turbulence blurs the
heavens so that no telescope, whatever its size, can resolve de-
tails smaller than something like 0.5 arcsecond, even on

MAX PLANCK INSTITUTE FOR ASTRONOMY (2)


mountaintops chosen for the best seeing. In terms of angular
resolution, the worlds biggest telescopes dont do much bet-

Right: The Calar Alto 3.5-meter telescope with the guidestar laser
(the narrower of the two black tubes) mounted on its side. Below:
The ALFA adaptive-optics system (red box) and an infrared camera
(golden canister) mounted at the focal plane of the Calar Alto reflec-
tor. ALFA was the first adaptive-optics system to use a laser guidestar
for regular astronomical observations.

32 October 2001 Sky & Telescope 2001 Sky Publishing Corp. All rights reserved.
EVERYONE AGREES THAT GROUND-BASED OPTICAL ASTRONOMY HAS NO ROOM TO GROW
WITHOUT ADAPTIVE OPTICS. SAYS FRANOIS RIGAUT OF THE GEMINI OBSERVATORY, WITHOUT
AO, BUILDING A NEW GENERATION OF VERY LARGE TELESCOPES MAKES NO SENSE AT ALL.
First focus: Distorted incoming
The idea harks back to the early 1950s, when Horace Bab- raw star images light wave
cock of the Hale Observatories in California first described the
principle. But it was beyond the technical possibilities of his
time. Only in the late 1980s was a prototype AO system in- Adaptive-Optics
stalled at the 3.6-meter telescope of the European Southern System
Observatory at Cerro La Silla in Chile. At that time, most de-
velopments in the field were being carried out in secret by
military agencies, mainly the U.S. Strategic Defense Initiative.
In May 1991, after the end of the Cold War, U.S. military AO High-speed
research was declassified and shared with astronomers world- tip-tilt mirror

or
at
wide. Over the past decade near-infrared AO systems have be-

tu
Ac
come almost a standard tool at many large observatories,
though using them is far from routine. And everyone agrees
that ground-based optical astronomy has no room to grow Collimator Telescope
Distorted lens
without adaptive optics. Says Franois Rigaut of the Gemini wavefronts

SKY & TELESCOPE DIAGRAM. STAR IMAGES: KECK OBSERVATORY


Observatory, Without AO, building a new generation of very Deformable Beamsplitter
large telescopes makes no sense at all. rubber mirror (dichroic)
Science
Infrared

Ac
camera

tu
Descrambling Starlight Flattened

at
or

Visible
So how does adaptive optics work? The basic idea is pretty wavefronts

s
simple. As with a coded TV signal coming through a descram-
bler, the blurred image of a star can be descrambled if you
know exactly how the scrambling was done. In other words, at Simple High-order
image-motion corrections
each instant you have to measure and correct for the effects of correction
all the atmospheric distortions inside a telescopes very long Wavefront-
line of sight. The distortions arise from tiny temperature dif- High-speed sensor
Corrected
processor camera
ferences among turbulent air pockets, or cells, ranging in size star image
from centimeters to meters. All the cells are rapidly blowing
across the telescopes light path, often in different directions at The basic parts of an adaptive-optics system were envisioned almost a
different altitudes. Each cell acts like a weak lens, slightly mis- half century ago, but it took decades to make them work. Light waves
directing the light rays passing through it. The result is a distorted by the atmosphere enter a telescope. Where they come to a
blurred and rapidly shivering image that typically changes focus, a star at high magnification appears as a distorted, rapidly shim-
hundreds of times per second. mering mess. (The five images here are freeze-frames of a seeing-
To make all the light rays from a distant star line up again, an distorted star, courtesy Keck Observatory.) Part of the light beam is di-
AO system must do two things: determine all the distortions at rected to a wavefront sensor that records the distortions. A high-speed
each moment, and impose their opposites on an optical element computer analyzes them and sends corrections to a rubber mirror,
in the telescopes light path. Both the angular size and the rapidi- which counter-distorts to cancel them out. The wavefront sensor then
ty of the distortions are worse at visual wavelengths than in the monitors the results. In systems that use a laser guidestar, a separate
near infrared; this is one principal reason that AO is pursued at tip-tilt mirror is also required to correct for overall image motion.
the longer near-infrared wavelengths. (Another is the fact that a
telescopes theoretical resolution is proportional to wavelength.) The optical element the rubber mirror is a thin
glass mirror whose surface can be slightly deformed into com-
plex shapes by tens or even hundreds of piezoelectric crystals
attached to its back. The crystals are tiny actuators: run a small
current through them, and they expand or contract a few mi-
JAMES R. GRAHAM / LICK OBSERVATORY / NSF

crons. If the tiny deformations of the rubber mirror match

A star before and after the new laser adaptive-optics system was
CENTER FOR ADAPTIVE OPTICS

turned on at Lick Observatorys 3-meter Shane reflector in California.


The infrared freeze-frame at left shows the all-too-familiar look of a
highly magnified star seen through the Earths atmosphere. When
the adaptive-optics system is switched on, the star image sharpens
nearly to the diffraction-limited size it would display if the atmos-
phere didnt exist.

2001 Sky Publishing Corp. All rights reserved. Sky & Telescope October 2001 33
AS SOON AS THE RUBBER MIRROR HAS THE CORRECT SHAPE, THE SENSOR MEASURES A FLAT
WAVEFRONT AND KNOWS THAT ALL IS WELL UNTIL THE NEXT MILLISECOND OR SO.

The asteroid 4 Vesta seen by the 10-meter Keck II tele-


scope before and after its adaptive-optics system was
turned on. Details as small as 0.05 arcsecond are re-
solved in the AO image, made at a wavelength of 1.6
microns. Courtesy Keck Adaptive Optics Group.

Strategy One: Fixing Wavefronts


Luckily, the same blurry star that youre trying to
focus can be used to measure the distortions that
have to be corrected, provided its bright enough.
Light waves from a star can be thought of as flat
planes before they hit the atmosphere and as dis-
torted, ripply surfaces by the time they enter the
telescope. The trick is to measure the distortions
(or, more correctly, counter-match) the distortions inflicted by of this wavefront all across the telescopes aperture and pass
the atmosphere, the blurring vanishes, and the telescope sees the information to the rubber mirror.
as if it were in outer space. To do this, a beamsplitter near the telescopes focus diverts
But this is easier said than done. The atmospheric distor- some of the starlight to a wavefront sensor. The most com-
tions have to be measured, and the required voltage for each monly used wavefront sensor (the Shack-Hartmann sensor,
actuator has to be calculated and applied, up to hundreds of named for its inventors) is an array of dozens of tiny lenses,
times per second. each with its own image detector. If the light beam had a flat

CULTIVATING AO: The Center for Adaptive Optics


Adaptive optics received a boost in No-
vember 1999 when the Center for Adap-
tive Optics (CfAO) opened at the Univer-
sity of California in Santa Cruz. As one of

AUSTIN ROORDA / NSF CENTER FOR ADAPTIVE OPTICS


the Science and Technology Centers
funded by the U.S. National Science
Foundation (NSF), the center will receive
$20 million over five years to bring AO
techniques to maturity. The center is di-
rected by astronomer Jerry Nelson and
sponsors a wide variety of astronomical
AO science projects.
Apart from developing new astro-
nomical techniques and instruments,
such as actuators and spectrographs, the CfAO also focuses Above: The living human retina, sharpened by adaptive optics in
on AO applications in vision science. Using an AO setup, oph- the lab. The image at left was photographed through the eyes
thalmologists have corrected for the complex, high-order cornea and lens, which have small, complex, normally uncor-
aberrations in human eyes. Researchers from the University of rectable errors. For the image at right, an adaptive-optics system
analyzed and compensated for these errors, allowing ophthalmolo-
Rochesters Center for Visual Science and the University of
gists to study individual cone cells clearly for the first time.
Houston have used the technique to image the living human
retina, seen through the eyes imperfect lens, sharply enough yield an ultraprecise map of corrections to be robotically
to easily resolve rod and cone cells and map their arrange- laser-sculpted into your cornea. The standard for fully cor-
ments. Test subjects looking outward through the system re- rected human vision could improve from 20/20 to 20/10,
port seeing more sharply than is possible in nature. Someday twice as sharp. The Web site for the Center for Adaptive Optics
adaptive-optics imagery of a guidestar on your retina may is at http://www.ucolick.org/~cfao/index.html.

34 October 2001 Sky & Telescope 2001 Sky Publishing Corp. All rights reserved.
A SHARPER SUN
At Sacramento Peak in New Mexico, the 76-centimeter
Dunn Solar Telescope operated by the National Solar Ob-
servatory (NSO) was recently outfitted with an adaptive-
optics system that provides extremely sharp images of
solar granulation. Although the technique differs from
classical AO (obviously, no guidestars are available), the
results are just as stunning. NSO director Stephen Keil says
a resolution of 0.15 arcsecond has been achieved, en-
abling astronomers to follow the evolution of very small
scale structures. By observing in a narrow wavelength
band, solar astronomers can see small magnetic flux tubes
penetrating the Suns surface.
The current AO system, which uses a 24-element Shack-

THOMAS RIMMELE / NOAO


Hartmann sensor, is the precursor for a much more versa-
tile system to be installed on the proposed 4-meter Ad-
vanced Technology Solar Telescope. The ATST, to be
constructed later this decade,has to have adaptive op-
tics, says Keil.

Above: Very fine structure exists between the granules (large, gray convection cells) covering the surface of the Sun, as seen in this image
made with an AO camera on the 76-centimeter R. B. Dunn Solar Telescope at Sacramento Peak, New Mexico. This 6-second exposure, made at
a wavelength of 430.5 nanometers in violet light, resolves details hardly more than 0.1 wide near its center. The bright points mark where
magnetic flux tubes emerge from the Suns interior. They can be studied spectroscopically only with the long exposures that adaptive optics
makes possible for very high resolution images.

wavefront, the light falling on each lenslet would be focused in apertures and detectors. Each detector has to be read out hun-
the center of each detector. If some parts of the beam are a lit- dreds of times per second. So unless the star is quite bright (for
tle tilted, the lenslets catching those parts will produce a fo- the AO system at the giant Keck telescopes this means magni-
cused spot displaced from the center. Measuring these dis- tude 13 or so the visual limit of a 6-inch telescope!), you
placements tells the shape of the wavefront. quickly run short of photons and theres nothing to measure.
This information is fed to a high-speed computer, which calcu- Does this mean adaptive optics can only be applied to bright,
lates the compensations to be sent to the rubber mirror. The rub- pointlike objects? Not if you use a simple trick. Suppose you
ber mirror is located ahead of the wavefront sensor in the optical want to observe a very faint galaxy. Its too dim and extended
train to create a closed feedback loop: as soon as the rubber mir- for the wavefront sensor to register. But if theres a bright star
ror has the correct shape, the sensor measures a flat wavefront close to the galaxy, you can use this as a guidestar. Provided
and knows that all is well until the next millisecond or so. the guidestar is within a few arcseconds of the galaxy, its light
Sounds easy. But wait a moment: it only works if the ob- passes through essentially the same air cells. Correct for one,
served object is bright enough. In the and you correct for the other and the telescopes main de-
wavefront sensor, the starlight is dis- tector (usually a sensitive infrared camera or spectrograph)
tributed across dozens of small lens will see the faint galaxy with unprecedented sharpness.

The moons of Jupiter and Saturn were practically featureless dots as


seen through Earth-based telescopes before the advent of adaptive
optics. These three-color infrared views were taken with the 10-
Europa meter Keck I telescope on December 13 and 14, 2000, in conditions
of mediocre seeing. Blue indicates wavelengths around 1.2 microns,
green 1.6 microns, and red about 2.2 microns. The smallest Galilean
satellite, Europa, is covered with relatively young ice, while its largest
sibling, Ganymede, is covered with mostly older ice. Titan is shrouded
by its own atmospheric haze at visible wavelengths, but the infrared
view here penetrates to the surface of Saturns largest moon. The
satellites are displayed at the same angular scale. Courtesy Keck
Ganymede 0.5 arcsecond Titan Adaptive Optics Group.

2001 Sky Publishing Corp. All rights reserved. Sky & Telescope October 2001 35
Strategy Two: Artificial Stars
The problem is that there arent enough 13th-magnitude stars
in the sky. Most interesting targets lack such a bright, close
neighbor. The fraction of the sky available to adaptive optics
the sky coverage, in AO parlance is thus pretty small, typical-
ly less than 1 percent. At near-infrared wavelengths the situation
is a little better, since the guidestar can be a few tens of arcsec-
onds away. Nevertheless, most of the sky remains inaccessible.
The solution is as straightforward as it is challenging. If
theres no guidestar where you want it, put one there yourself.
If you really want to go faint and have a large sky coverage,
you need a laser guidestar, says Geminis Rigaut.
Laser-beacon adaptive optics were developed in the 1980s by
Robert Fugate and his collaborators at the U.S. Air Forces
Starfire Optical Range in New Mexico (S&T: May 1994, page
24, and June 1994, page 20). By using a powerful laser beam
tuned to a wavelength of 589 nanometers, the absorption and
emission wavelength of sodium, telescope operators can make
atoms of sodium near the top of the atmosphere glow. If a 20-

LAWRENCE LIVERMORE NATIONAL LABORATORY / LICK OBSERVATORY (2)


watt laser is focused tightly enough, the result is an artificial,
yellow-orange star glowing at an altitude of some 90 kilome-
ters, the height of a particularly sodium-rich layer. This laser
guidestar is too faint to be seen by the unaided eye, but its
bright enough for the wavefront sensor. The laser beam doesnt
interfere with the observations because it is confined to a sin-
gle wavelength in the visible spectrum, while most adaptive-
optics observations today are done in the infrared.
But now matters really get complicated. First of all, its no
mean feat to build the specialized 20-watt laser. High-
performance, high-quality lasers are still very expensive, says
Rigaut. Acton at Keck Observatory agrees. To get this amount
of power, you have to push everything to the very limit, he
says. For instance, it takes a staggering 50,000 watts of power
input to get the desired laser output. Tuning lasers to the cor- Above and below: The newest laser AO system being tested at Lick Ob-
rect sodium wavelength involves a challenging technique that servatorys 3-meter Shane reflector on Mount Hamilton in California.

36 October 2001 Sky & Telescope 2001 Sky Publishing Corp. All rights reserved.
THE SOLUTION IS AS STRAIGHTFORWARD AS IT IS CHALLENGING. IF THERES NO GUIDESTAR
WHERE YOU WANT IT, PUT ONE THERE YOURSELF. IF YOU REALLY WANT TO GO FAINT AND
HAVE A LARGE SKY COVERAGE, YOU NEED A LASER GUIDESTAR, SAYS GEMINIS RIGAUT.

uses frequency doublers, dyes of liquid ethanol containing or- Fortunately this star can be pretty dim, since its light doesnt
ganic molecules, and dye-master oscillators. Its a miracle of have to be divided into many pieces by the wavefront sensor.
physics that I still have to understand, says Acton. Adaptive optics doesnt look so simple with all these com-
Second, the powerful laser could blind airline pilots and plications. Apart from a wavefront sensor, a high-speed com-
damage detectors on military spy satellites. Astronomers using puter, and a deformable mirror, you now also need a very
adaptive-optics lasers run an infrared camera in tandem that powerful and tricky laser, a camera to spot airplanes, a hotline
automatically shuts the system down when a plane comes by. to NORAD, and a separate aperture-tilt correction loop.
All observing programs, including detailed timing schedules, Little wonder that laser-beacon AO systems are still pretty
have to be submitted to the North American Aerospace De- much in an experimental phase, despite many years of develop-
fense Command (NORAD), which gives the observatory a list ment. Not counting a few military sites, there are just two gen-
of time slots when the laser has to be turned off because a spy uinely working systems. The first was ALFA (Adaptive optics
satellite is passing near the field of view. with a Laser For Astronomy), installed at the 3.5-meter telescope
Last but not least, though a laser guidestar can be used to of Calar Alto Observatory in southern Spain. The second, based
measure the curvature of the wavefront distortions, it cannot on technology developed at Lawrence Livermore National Labo-
be used to measure the images so-called jitter its bulk mo- ratory, has been set up at the 3-meter Shane reflector of Lick
tion from side to side, an effect known as full-aperture tilt. The Observatory at Mount Hamilton, California. Both systems have
problem is that the same large-scale air cells affect the out- had their growing pains. The Lick system is only now being used
going laser beam and the incoming starlight. To measure the for its first science observations. At Keck Observatory, Acton and
full-aperture tilt, its still necessary to use a natural guidestar. his collaborators are installing a laser system of their own.

RUBBER SECONDARIES
All current adaptive-optics systems use a small flexible mirror those in audio speakers; its shape will be controlled by the
to correct the wavefront of the incoming light. Because its same AO feedback loop that normally controls the small rub-
small it has to be placed close to the focus in its own separate ber mirror. With the active mirror part of the telescopes main
equipment package. Of course it would be more straightfor- optics, AO observations should become much more routine.
ward to adjust the shape of the telescopes primary mirror, At the Large Binocular Telescope (LBT) being built on Ari-
but its size and stiffness would make this very difficult. How- zonas Mount Graham, both 8.4-meter units will have adaptive
ever, astronomers and engineers at the University of Arizona secondary mirrors, says Close.This will enable us to use them
in Tucson have developed an AO system that uses a tele- as an interferometer without using any additional optics.
scopes secondary as the rubber mirror.Its working! says Ari-
zonas AO wizard Laird Close.
According to Close, current AO systems lose a great deal of
light because of the many additional mirrors involved in the
packages. Also, the mirrors inevitably emit a little heat, de-
grading infrared observations. Finally, each additional optical
STEWARD OBSERVATORY / CENTER FOR ASTRONOMICAL ADAPTIVE OPTICS

element creates additional scattered light, further decreasing


the signal-to-noise ratio.
A flexible secondary mirror is being installed in the Multi-
ple Mirror Telescope (MMT) at Mount Hopkins near Tucson
(which, despite its name, has been outfitted with a single, 6.5-
meter primary). The new secondary is 60 centimeters in diam-
eter but a mere 2 millimeters thick. It weighs less than a kilo-
gram. The flexible secondary uses 336 voice coils similar to

The first rubber secondary mirror integrated into a large telescope.


The 6.5-meter MMT on Mount Hopkins, Arizona, will use this ultra-
thin, 60-cm secondary backed by 336 magnetic actuators (visible
through the glass because it had not yet been coated).

2001 Sky Publishing Corp. All rights reserved. Sky & Telescope October 2001 37
1999 NEELON CRAWFORD, POLAR FINE ARTS / GEMINI OBSERVATORY / NSF (2)

The new 8.1-meter Gemini North telescope on Mauna Kea began sci-
ence operations a year ago; its twin at Cerro Pachn, Chile, should
start work late this year. Adaptive optics will be an integral feature of
both. For now, Gemini North is equipped with a curvature AO system,
named Hokupaa, on loan from the University of Hawaii Institute for
Astronomy. These outside and inside views show Gemini North with
its side ventilation vents open at sunset.

Strategy Three: Curvature Systems


Franois Roddier of the AO group at the University of Hawaii
is convinced that laser-beacon systems will someday be regard-
ed as the dinosaurs of adaptive optics: too large, too ineffi-
cient, and bound for extinction. My hope is that we will never
have to use laser guidestars anymore, says Roddier.
His alternative is a different type of wavefront sensor without
the disadvantages of the classical Shack-Hartmann design. In-
stead of distributing the guidestars light across many tiny aper- cy of the actuators in a curvature system is much higher than
tures, a single curvature sensor, as it is called, uses all of the in a traditional system, because the shape corrections can be
image to measure the shape of the wavefront. The details are applied to the rubber mirror with fewer actuators. The reason
hard to grasp, but the principle is pretty simple. Two defocused is that the desired shape is not described as a large number of
images of the guidestar are taken, one on each side of the focal offsets at particular points but as a relatively small number of
plane. Amateurs who use the classic star test to judge backyard superimposed curvatures.
telescopes will know whats coming next. If the wavefront were The successful AO system currently being used at the Gemi-
perfectly flat, the brightness distribution in the two images ni North telescope, called Hokupaa (Hawaiian for immov-
would be identical. Any distortions in the wavefront make them able star or Pole Star), is a curvature system originally built
look different. By precisely comparing the two extrafocal im- by Roddiers group at the Institute for Astronomy of the Uni-
ages, the system can reconstruct the wavefronts distortion. versity of Hawaii. The system, installed in May 1999, uses just
Curvature systems have a limiting magnitude of 15 or 16, 36 actuators. Roddier admits that Hokupaa is not yet as good
says Roddier, corresponding to a sky coverage of 10 percent. as the AO system at Keck, but in early 2002 it will be upgraded
Future systems may go fainter. Moreover, he adds, the efficien- to a system with 85 actuators. Hokupaa-85 will do as well as

38 October 2001 Sky & Telescope 2001 Sky Publishing Corp. All rights reserved.
GEMINI OBS. / NSF / UNIVERSITY OF HAWAII INSTITUTE FOR ASTRONOMY / DANIEL POTTER
GEMINI OBSERVATORY / NSF

Above: Pluto and its satellite, Charon, never appear more than 0.9 arcsecond apart, but the
Gemini North telescope easily resolved them with the Hokupaa adaptive-optics system on
several nights in June 1999. Charon orbits Pluto every 6.4 days. Both appear slightly larger
than point sources in these infrared images.

Right: Without adaptive optics the young binary star GG Tauri would be a featureless fuzzball
as seen from Earth. This near-infrared view was obtained by the Hokupaa system on Gemini
North last February 24th. A massive ring of dust with a ragged inner edge orbits about 140 as-
tronomical units from the star pair (masked out and indicated by symbols) 450 light-years
away. The stars have a total mass of 1.2 Suns; the ring has about 0.3 solar mass. This image is
1
sharper than Hubble Space Telescope views of GG Tauri.

CURVATURE SYSTEMS HAVE A NUMBER OF SHORTCOMINGS, SAYS RIGAUT, WHICH MAKE


THEM ILL SUITED FOR REALLY LARGE TELESCOPES. BUT THE UNIVERSITY OF HAWAIIS
FRANOIS RODDIER DISAGREES. WE ARE TRYING TO PROVE THE OPPOSITE.

a conventional system with a few hundred actuators, but its mapping many cross sections. MCAO enables astronomers to
much cheaper, he says. The Shack-Hartmann AO system at build a three-dimensional model of atmospheric turbulence
Keck II uses 349 actuators. all along a telescopes light path, not just a two-dimensional
At the neighboring 8.3-meter Subaru telescope, Japanese as- map of its effect at the image plane.
tronomers have set up a 36-actuator curvature AO system For large telescopes this becomes more and more impor-
comparable to Hokupaa. According to Subarus instrument tant, as a simple argument shows. Suppose you had a 30-meter
group manager Takuya Yamashita, it will be a competitor for telescope aimed at a galaxy. The light from the galaxy collected
both the Gemini and Keck systems. First light was achieved in by the giant mirror has traversed a cylinder of air 30 meters in
late 2000. And at the European Southern Observatorys Very diameter. However, the light from the laser guidestar, which
Large Telescope at Cerro Paranal, Chile, a number of curva- sits some 90 kilometers up, does not traverse this entire cylin-
ture AO systems are being planned, in addition to a 196- der of air. Instead, it traces out a cone with a base diameter of
actuator Shack-Hartmann system (named NAOS/Conica) now 30 meters tapering to a point at the top. Turbulent air within
being installed. the cylinder but outside the cone will go unmapped, and the
Rigaut and his Gemini colleague Brent Ellerbroek say that galaxy cannot be fully unblurred.
even though a duplicate of Hokupaa-85 will be installed on Multi-conjugate adaptive optics solves this problem by
Gemini South at Cerro Pachn in Chile, the next-generation using multiple guidestars, wavefront sensors, and rubber mir-
AO system at Gemini, called Altair, will be of the classical rors. It also overcomes the severe limits on the size of the
Shack-Hartmann type. Altair, built at the Herzberg Institute of sharpened field of view. Jacques Beckers (National Optical As-
Astrophysics in Victoria, British Columbia, will exploit a laser tronomy Observatories) was the first to describe this tech-
system and is expected to be installed at Mauna Kea in 2002. nique as atmospheric tomography (in 1988), though it had
Curvature systems have a number of shortcomings, says been discussed in the literature before. Ellerbroek is one of the
Rigaut, which make them ill suited for really large telescopes. few people who developed the technique during recent years.
But Roddier disagrees. Many people still seem to think that, Together with Rigaut he is designing an MCAO system for
he comments. We are trying to prove the opposite. Gemini that could be installed four years from now. The idea
is to use five laser guidestars arranged around the field of view.
Strategy Four: Atmospheric Tomography We have a lot of confidence, he says. The basic components
If astronomers plan to build really large telescopes, with aper- are the same [as for a classical laser AO system]. We dont see
tures of 20 meters or more, current techniques using one any physical uncertainties.
guidestar (laser or otherwise) and one rubber mirror will not So how is the three-dimensional tomography achieved? Be-
suffice. Instead, astronomers will have to rely on a revolutionary cause the cones are oriented differently, the altitude of a piece
technique called multi-conjugate adaptive optics, or MCAO. of turbulence thats shared by two or more of them can be
This technique has been described as atmospheric tomogra- found essentially by parallax. Five laser guidestars should be
phy. Its reminiscent of medical tomography, whereby a sur- enough to construct a full 3-D map of the turbulence above
geon obtains a three-dimensional view of a human body by the telescope. Of course, this requires very fast computers,

2001 Sky Publishing Corp. All rights reserved. Sky & Telescope October 2001 39
Two AO views of the Milky Ways crowded center.
The field at left is an uncompensated infrared view
made by Gemini North in excellent (0.65-arcsecond)
seeing. The Hokupaa adaptive-optics system dra-
matically increased the resolution, as shown in the
upper inset. The lower inset shows a neighboring
field imaged with the Pueo
AO system on the 3.6-meter
Canada-France-Hawaii Tele-
scope. The massive but in-
active black hole at the
heart of the galaxy lies
approximately at the cen-
ter of the Pueo star field.
Wide-field image and
upper inset: Gemini
Observatory/NSF/Uni-
versity of Hawaii Insti-
tute for Astronomy/
Ruth A. Kneale. Lower
10
inset: Canada-France-
Hawaii Telescope.

FOR THE NEXT 10 YEARS OR SO, THERES NO WAY TO BEAT . . . THE NEXT GENERATION
SPACE TELESCOPE AT VISIBLE WAVELENGTHS, SAYS GEMINIS BRENT ELLERBROEK. BUT
WE CERTAINLY CAN BE COMPETITIVE WITH NGST IN THE NEAR INFRARED.

since the tomography calculations have to be executed many afford to spend much money on building a prototype first.
hundreds of times per second. Money will certainly be a key issue. An MCAO system needs
In the system Ellerbroek and his colleagues are designing, more wavefront sensors, rubber mirrors, electronics, computer
three rubber mirrors compensate separately for turbulence in power, and last but not least, five times as many lasers as a con-
three altitude zones: one close to the telescope, one about 4 ventional laser-beacon system. But thats unless lasers can be
kilometers up, and one about 8 kilometers up. These altitudes discarded altogether in favor of curvature systems, says Roddier.
can be adjusted to match wherever most of the turbulence is If the field of view can be increased to a few arcminutes, it may
happening on a given night. In technical jargon, the rubber contain enough natural guidestars at the limiting magnitude for
mirrors are optically conjugated to specific altitudes; hence a curvature system to do multi-conjugate adaptive optics. No
the name multi-conjugate adaptive optics. need for laser guidestars anymore thats my hope.
Computer simulations have indicated the great potential of Less than 20 years after its inception, adaptive optics has
this technique, and Ellerbroek and Rigaut say MCAO will keep evolved into a mature, albeit extremely challenging, technique.
Gemini competitive with other ground- and space-based tele- It stands at the threshold of becoming a routine observing
scopes well into the 21st century. Initially, of course, most ob- mode at many large telescopes. Twenty years from now every
servations will be carried out in the near infrared, since every professional telescope will probably use some sort of AO sys-
type of adaptive optics is easier to do at longer wavelengths. tem, and observing volcanic eruptions on Io from a moun-
For the next 10 years or so, theres no way to beat the ex- taintop on Earth will no longer be considered extraordinary.
pected performance of the Next Generation Space Telescope at For some time to come, however, each AO result, especially
visible wavelengths, says Ellerbroek, referring to the proposed from telescopes as large as Gemini and Keck, will be consid-
6-meter successor to the Hubble Space Telescope (recently ered a new treasure. Theres so much to find out there, says
downsized from 8 meters). But we certainly can be competi- Acton, and were lucky to be in a position to collect the most
tive with NGST in the near infrared. exciting results. I really feel like the first kid in the garden dur-
Other groups are also studying MCAO, including the teams ing an Easter-egg hunt.
designing the next generation of very large ground-based tele-
scopes. Building extremely large telescopes on Earth makes no Contributing editor Govert Schilling writes about astronomy from
sense at all without MCAO, says Ellerbroek. The European his hometown of Utrecht, the Netherlands, though his passion for large
Southern Observatory is working on a demonstration system telescopes (and volcanoes) has taken him to Hawaii and Chile. His
that will probably be completed before the Gemini system. book Flash! The Hunt for Cosmic Super Explosions, about gamma-
They are in a luxurious position, comments Rigaut. We cant ray bursts, will soon be published by Cambridge University Press.

40 October 2001 Sky & Telescope 2001 Sky Publishing Corp. All rights reserved.

Das könnte Ihnen auch gefallen