ÐÓ°ÉÔ­´´

In late 1993, if all goes according to schedule, a space shuttle loaded
with spare parts will dock with the Hubble Space Telescope. Astronauts will
replace a defective gyroscope and, especially, the existing solar panels,
whose vibrations shake the entire craft and prevent it from taking accurate
bearings of its position. They will also replace the main imaging instrument,
the wide-field/planetary camera, with a new version that compensates for
the focusing problems of the telescope’s 2.4-metre main mirror. But that
is as far as NASA has got with its plans. The agency has still not decided
whether to correct the vision of the telescope’s other optical instruments,
or to wait three years and replace them with new ones. Next month it expects
to receive a detailed design of the system needed to correct the optics
and to have made up its mind what to do by May.

The planned repair mission is one sign of progress for the space telescope.
Another is the emergence of the first scientific results: computer processing
has generated some impressively sharp images despite the flawed main mirror
and imprecise pointing . But major challenges remain. Image enhancement
works only for bright objects, and even then there are serious limitations.
Focusing and pointing problems make some observations impossible and lengthen
the time needed for others. They also reduce the time the telescope is usable.
Furthermore, engineers have yet to identify all the problems precisely enough
to fix them, and NASA management is hesitant about building a proposed corrective
optics module (This Week, New ÐÓ°ÉÔ­´´, 23 February). Even when it is fixed,
Hubble will not perform as well as it was supposed to. And experience teaches
that NASA schedules, like that of Hubble itself, tend to slip.

Hubble was designed to see to the edge of the Universe. The 2.4-metre
primary mirror collects light and directs it onto a 0.3-metre secondary
mirror, which focuses the beam back through a central 0.6-metre hole in
the primary mirror to five instruments. The telescope collects light from
a region of the sky 28 arcminutes across-a field of view roughly equivalent
to the diameter of a full Moon seen from Earth-and projects it onto an area
about the size of a dinner plate in the focal plane 1.5 metres behind the
primary mirror.

No single instrument sees the entire field of view. A small flat mirror
in front of the focal plane directs the central 160 arcseconds to the wide-field/planetary
camera, mounted at one side of the telescope. Separate fixed apertures in
the focal plane collect light for the four other instruments; the faint
object camera, the Goddard high-resolution spectrograph, the faint object
spectrograph and the high-speed photometer. These four instruments are mounted
behind the focal plane around the axis of the telescope, each in a box about
3 metres high and 1.5 metres across.

Each instrument has a distinct purpose. The wide-field/planetary camera
contains two sets of optics: one records images of distant objects across
a wide field of view and the other, with higher magnification, operates
across a narrower field of view to record images of planets, which are much
nearer and brighter. The faint object camera also records images, but is
more sensitive to blue and ultraviolet light, and provides more detail than
the wide-field/planetary camera. The two spectrographs analyse the wavelengths
of light from different types of objects: the faint object spectrograph
retrieves as much information as possible from faint objects while the Goddard
high-resolution spectrograph studies brighter ones in detail in the ultraviolet
range. The high-speed photometer is designed to measure stellar brightness
with extreme precision.

Hubble was launched last April and sent back its first images a month
later. The images were out of focus and scientists soon discovered that,
despite checks throughout its manufacture, the primary mirror had been polished
to the wrong shape. Its edge is 0.002 millimetres too low, so light reflected
from that area is focused 38 millimetres behind light reflected from the
region around the central hole of the concave mirror.

This defect, called spherical aberration, has a dramatic effect on image
quality. Hubble was supposed to focus 70 per cent of starlight onto a 0.1-arcsecond
spot. At the compromise focal point, 7 millimetres behind the focus of the
central region, and 31 millimetres from the focus of the mirror edge, only
15 per cent of light is concentrated in the spot. The rest is spread in
a complex ‘halo’ three arcseconds across.

The spherical aberration has not kept Hubble from taking a few impressive
pictures of bright objects, using light from the central spots. Even without
image processing, the faint object camera clearly resolved Pluto and its
moon Charon when they were only 0.9 arcseconds apart, something impossible
from the ground. However, Duccio Macchetto, project scientist for the faint
object camera at the Space Telescope Science Institute, says the aberration
causes images to overlap and reduces the instrument’s sensitivity by a factor
of 6 to 16, or 2 to 3 magnitudes. (The magnitude scale is an arbitrary measure
of the brightness of celestial objects. A magnitude difference of 1 unit
corresponds to a brightness ratio of 2.512; a difference of 5 units corresponds
to a brightness ratio of 100. The larger the unit of magnitude, the fainter
the object.) The Goddard high-resolution spectrograph ‘can do most of the
science’ planned, says Ken Carpenter, a member of the instrument team at
the NASA Goddard Space Flight Center, but the observations take up to five
times longer than expected because light intensity is reduced. Poor image
quality ‘wreaks havoc’ in analysing data from the wide-field/planetary camera,
complains principal investigator James Westphal of the California Institute
of Technology.

Compensating for the flawed mirror, like correcting imperfect vision,
requires precise measurement of the defect, but that has proved difficult.
The spherical aberration makes it very hard to align the telescope, and
could hide other optical defects. The real error may not be determined until
astronauts visit Hubble in 1993.

Good observations with any telescope take time. The instrument must
point accurately at the desired object, and track it precisely as the telescope
(or, for a terrestrial telescope, the ground underneath it) moves. Small
errors cause blurred images; larger errors may make the telescope miss the
target altogether. Hubble must be pointed extremely accurately to achieve
the promised high image quality and to direct light to the tiny apertures
of some of the scientific instruments. However, astronomers have been frustrated
by vibrations from the solar panels and by the effects of optical flaws
on the sensors that orientate the telescope in space.

The largest vibrations occur when the satellite passes into and out
of the Earth’s shadow. The solar panels bend as the temperature changes
from -90 °C in the Earth’s shadow to +100 °C in sunlight. The European
Space Agency and British Aerospace, which designed and built the panels,
had expected the change to take between 1 and 5 minutes; in fact, it takes
only 11 seconds, which causes the telescope to shake so much that it can
lose track of the stars that it is using for guidance. Joints to accommodate
differences in thermal expansion across the panels cause smaller oscillations.
The vibrations ‘don’t ruin data, you just lose some of the light’ in the
spectrographs, says Ed Weiler, a scientist working on the Hubble project
at NASA headquarters. They are more of a problem for the cameras, he admits,
particularly when they are taking long exposures of faint objects.

Even without the vibrations, Hubble has a hard time tracking stars with
the desired accuracy of 0.020 arcseconds in a ‘coarse’ mode (for general
observations) and 0.007 arcseconds in a ‘fine’ mode (when extreme accuracy
is needed). Tracking in both modes relies on three arc-shaped sensors, which
are mounted around the edge of the focal plane and use the known positions
of pairs of stars for celestial triangulation. The sensors were supposed
to work with stars as faint as magnitude 14.5, but because the mirror’s
optical aberration diffuses starlight, they cannot guide the telescope reliably
on stars fainter than magnitude 13 or 13.5. This means that there are fewer
stars that Hubble can use to track the most distant and interesting objects,
which, because the spacecraft must be orientated so that the solar panels
face the Sun, restricts many observations to certain times of the year.

The wide-field/planetary camera can tolerate pointing errors because
the desired object will still fall into its wide imaging area; the spectrographs,
with their tiny apertures, cannot. According to Vesa Junkkarinen, a research
associate at the University of California at San Diego, the 4.3-arcsecond
aperture of the faint object spectrograph has missed many targets and the
0.3-arcsecond aperture of the high-resolution spectroscope is virtually
unusable. Even worse off is the high-speed photometer, which will be removed
from Hubble if NASA goes ahead with a proposal to add a module to correct
spherical aberration in 1993.

Until Hubble is repaired, scientists must rely on computers to enhance
the images they receive. To do this, scientists first measure the pattern
the defective optics form when viewing a point source, such as a star. Using
that pattern, a computer can work backwards from a distorted image to estimate
what the image would have looked like through perfect optics. Computer enhancement
can greatly clarify images of bright objects in a dark field, but it cannot
do anything for faint objects on the edge of the Universe. There is simply
not enough light available to enhance their images, says Weiler.

Even using a large number of processors working in parallel, there are
significant limitations to the technique. Bright stars saturate Hubble’s
electronic detectors, which then underestimate intensity and prevent images
from being completely corrected. For diffuse objects, such as gas clouds,
the detectors have trouble distinguishing between unfocused and focused
light. Moreover, the distortion produced by the defective optics varies
across the image plane, which means that a range of patterns is needed to
correct the image. Some astronomers simply do not trust image processing-computers
can interpret background radiation as stars, complains Westphal. Others
want to study objects that cannot be clarified. Image enhancement is essentially
a stopgap measure until the telescope can be fixed.

With this in mind, NASA is now deciding what to do about Hubble. The
agency has already decided to replace the wide-field/planetary camera in
1993; it has two choices about what to do about the other instruments. The
Hubble Space Telescope Strategy Panel, which the agency appointed to come
up with a solution, wants the astronauts on the 1993 mission to replace
the seriously impaired high-speed photometer with a set of optics that would
let the two spectrographs and the faint object camera function with the
flawed mirror. Alternatively, NASA could make do with these intruments until
1996 when it would be in a position to install a new generation of equipment
that would properly focus light from the flawed mirror.

The first option involves the deployment of an optical module called
COSTAR, standing for Corrective Optics Space Telescope Axial Replacement
(This Week, New ÐÓ°ÉÔ­´´, 12 January). COSTAR would fit Hubble with corrective
optics that are analogous to spectacles, but made of concave mirrors rather
than lenses. Five pairs of mirrors 10 to 30 millimetres in diameter would
be mounted inside a box the same size as the high-speed photometer. After
astronauts mount COSTAR, arms in the box would move the optics into place
in front of the image plane. One mirror in each pair would be optically
perfect, while the other would be made with exactly the amount of spherical
aberration needed to cancel the flaw of the primary mirror. The corrective
mirror in each pair would lock into a fixed position, but the position of
the perfect mirror would be adjustable over tens of micrometres. The corrective
mirrors would have a complex, aspherical and asymmetrical shape. Such mirrors
are far from easy to manufacture, but the strategy panel called them ‘well
within the state of the art’.

COSTAR designers picked small mirrors because they do not intend to
correct the entire field of view. Instead, they plan to correct only the
regions where apertures collect light for the axial scientific instruments,
says James Crocker of the Space Telescope Science Institute and a member
of the strategy panel. One pair of mirrors would suffice for the high-resolution
spectrograph; the faint object spectrograph and faint object camera need
two pairs each. Making such small mirrors would be faster, easier and much
less costly than making a single large mirror to correct the entire field
of view. NASA remains unconvinced, however.

The agency’s dilemma is that COSTAR is only a temporary fix for three
instruments now aboard Hubble. The two spectrographs are expected to be
replaced in 1996 by new instruments with internal corrective optics, leaving
only the faint object camera in need of COSTAR. ‘The question we’re trading
off is if we should spend time and money on COSTAR to correct only first
generation instruments, or live with what we’ve got and procure the new
instruments as fast as we can,’ said Dennis McCarthy, deputy manager of
the Hubble programme at NASA headquarters in Washington. The decision is
crucial to the European Space Agency (ESA), which supplied the faint object
camera and regards COSTAR as the only practical way of rehabilitating its
instrument. ‘We’ve looked at various options, including opening up the faint
object camera in orbit, but they were not considered good solutions,’ said
Robin Laurence, Hubble project manager at ESA. Unlike the other instruments,
the faint object camera was not due for replacement in 1996. If NASA abandons
COSTAR, Laurence concedes that ESA will have to consider building a new
instrument. He added that NASA ‘has to consider the ESA contribution’ as
well as its own trade-offs.

NASA had long planned to exchange instruments attached to Hubble, and
the long delay in its launch-at one point planned for 1985-added impetus
because new technology became available for use in new instruments. Even
before Hubble was launched last April, the Jet Propulsion Laboratory was
working on an upgraded wide-field/planetary camera. Originally intended
as a backup, the new version includes detectors that will offer higher resolution
and better sensitivity, and has been modified to reduce contamination, apparently
by organic materials, that has reduced the ultraviolet sensitivity of its
predecessor. Slight changes in the optics that focus light onto the detector
arrays should correct the new version for spherical aberration. According
to the strategy panel, the main challenge will be to align the correcting
mirrors with a lateral tolerance of about 1 per cent, or within about 60
micrometres. For reasons that remain unclear, mirrors of the first camera
are apparently misaligned by between 5 to 10 per cent. Because of the tight
tolerances, the new camera may require internal adjustments to align the
optics in space.

But even with the correction for Hubble’s defective optics, the new
camera will not be as good as it would have been without spherical aberration.
Like its predecessor, the new wide-field/planetary camera uses a four-faceted
prism that rotates to switch light between its two internal modules, the
wide-field camera and the planetary camera. Both modules of the camera contain
four imaging systems. Each of these receives light from one prism facet
and directs the light to separate arrays of charge-coupled-device detectors,
which form the four quadrants of an image. Light from the telescope will
be out of focus when it strikes the prism, so light from one object may
be split between two prism facets, producing extra images and stray light.

The new version of the camera will be installed on the 1993 mission
to Hubble. NASA is developing two more instruments to be installed in 1996:
the first is the near-infrared camera and multi-object spectrograph (NICMOS);
the second is the space telescope imaging spectrograph (STIS). Both will
include internal optics to correct for spherical aberration. NICMOS will
be the first Hubble instrument sensitive in the near infrared, recording
images and spectra at wavelengths of 1 to 3 micrometres. Three internal
cameras will image different areas of the Hubble focal plane onto infrared
detector arrays. Three separate spectrographs will cover the same wavelength
range with high resolution. STIS will operate from 105 to 1100 nanometres,
a wider range than the two current spectrographs combined, and will replace
both. It will also have better sensitivity, better ultraviolet optics, and
less scattered light. If uncertainties in the measured spherical aberration
are not resolved, the instrument will include a carousel of several correcting
mirrors with slightly different shapes. Operators on the ground will select
the mirror which gives the best focus, like an optician testing a person’s
vision.

Deploying COSTAR will not help the guidance sensors to point Hubble
in the right direction: in fact, if COSTAR is to be a success, NASA must
first improve the performance of the sensors by resetting them for an optical
system with spherical aberration. The strategy panel says this should be
possible by revising the software that controls the sensors’ operation.

Further work is needed to calibrate the locations of the instrument
apertures on the image plane relative to the guidance sensors. Those positions
have not been known well enough to direct light reliably into instrument
apertures. Calibration was almost complete at the end of January, but it
may remain difficult to use the smaller apertures.

To overcome the vibration problem, NASA programmed Hubble’s computers
to induce offsetting motion to compensate for the 11-second vibrations of
the solar panels. However, this solution had to be abandoned because it
used too much of the on-board computer’s capacity, says Weiler. Although
the agency is going to try again with a temporary ‘software fix’, it is
pinning its hopes for a permanent solution on new solar panels that do not
suffer uneven thermal expansion.

NASA plans to install the new arrays in 1993, and ESA and British Aerospace
are studying ways to control vibrations. ‘Replacement of the solar arrays
is not an extra cost; it was already in our plan,’ said Laurence of ESA.
Solar cells degrade after long exposure to space, and NASA had planned to
install new panels in 1996. British Aerospace began working on the second
solar array before Hubble was launched to save production costs, originally
planning to store the array. That head start will help the company meet
the 1993 deadline.

The strategy panel has already sketched out plans for the 1993 repair
mission. Two astronauts would work in space for two or three six-hour shifts.
In the first, they would replace the defective gyroscope, replace the wide-field/planetary
camera with the upgraded version, and replace the high-speed photometer
with COSTAR if NASA goes ahead with that program. If COSTAR is installed,
they would check that it was working properly. In their second shift, the
astronauts would remove the old solar panels and replace them with the new
set. Ever-cautious planners reserved time for a third work period in case
astronauts cannot complete their tasks in the first two.

The need to wait three years before Hubble can be fixed-and even longer
until the new instruments are properly calibrated-frustrates many astronomers.
More than a decade has passed since engineers began developing instruments
for Hubble, but astronomers have yet to get the spectacular results they
had hoped for. The few results that emerged in January at an American Astronomical
Society meeting in Philadelphia were collected while the telescope was being
tested and calibrated. Operation was still restricted at the end of January,
although Crocker said that ‘we’ve actually begun the transition into more
routine operations,’ including regular observations.

Privately, some astronomers expressed bitterness and anger as well as
frustration at the fate of what former NASA administrator James Beggs in
1982 said might be ‘the most important scientific instrument ever flown’.
Hubble was designed to probe the fringes of the Universe, and observe individual
stars in other galaxies, including the Cepheids, which are bright stars
whose regular variations would have let astronomers measure intergalactic
distances accurately for the first time. One astronomer caustically suggested
taping the title ‘Cepheid variables in the Virgo Cluster’ above a blank
poster as a reminder that Hubble will probably never make the spectacular
observations once expected of it.

* * *

The scientific results that beat anything seen from Earth

Astronomers had ambitious plans for the Hubble Space Telescope. Before
launch, they talked of probing galaxies and quasars at the edge of the observable
Universe, up to 15 billion light years away. They expected Hubble would
have much higher resolution and be able to detect much fainter objects than
ground-based telescopes. They planned to calibrate intergalactic distances
by observing bright variable stars in the Virgo Cluster, 60 million light
years away.

The optical and pointing problems discovered after launch have put many
of those plans on hold, until Hubble can be repaired, and may prevent others
altogether. Most Hubble observations so far have been intended primarily
to diagnose its problems. However, some results have begun to emerge, including
clearer pictures and more detailed spectra of comparatively bright objects
than could be obtained from the ground. It will take some time for astronomers
to analyse that data.

The faint object camera clearly resolved Pluto and its moon Charon when
they were 0.9 arcseconds apart. The moon was not discovered until 1978 because
the atmosphere blurs images so much that ground-based telescopes can barely
separate the moon from the planet. The Hubble image shows that Charon is
much fainter than the larger Pluto. This supports the suggestion that ‘water
ice’ covers Charon, while brighter ‘methane ice’ covers Pluto, says Duccio
Macchetto, project scientist for the faint object camera at the Space Telescope
Science Institute. Observations at other wavelengths should tell astronomers
more about the compositions of Pluto and Charon.

Hubble observations of a ring around Supernova 1987A in the Large Magellanic
Cloud have let astronomers measure its distance at 169 000 light years,
with only a 5 per cent margin of error. The faint object camera shows that
the structure is a real ring, not a side view of a gas shell, says Nino
Panagia, a member of the ESA team working on the faint object camera at
the Space Telescope Science Institute. Measuring the ring’s diameter, its
geometry, and the time reflected light from the supernova took to reach
Earth made it possible to calculate the physical size of the ring and its
distance away.

The Goddard high-resolution spectro-graph has revealed details in the
ultraviolet spectrum of Aldebran not seen before. Ken Carpenter, a member
of the instrument team at the NASA Goddard Space Flight Center, says the
Hubble data contains less noise than the spectra collected previously.

Sally Heap of the NASA Goddard centre, co-principal investigator for
the high-resolution spectrograph, studied Melnick 42, a star in the Large
Magellanic Cloud that is more than a million times as luminous as the Sun
and has a mass 80 to 100 times as great. One of the most massive stars known,
it is only about two million years old, and is expected to explode as a
supernova in a few million years. Its ultraviolet spectrum shows a strong
stellar wind, moving up to 2900 kilometres per second, which ejects one
solar mass every 100 000 years. The new observations are helping astronomers
understand how stellar evolution depends on the initial proportion of heavy
elements, which differs in our galaxy and in the Large Magellanic Cloud.

The wide-field/planetary camera has studied the odd-shaped nebula around
another star nearly 100 times as massiveas the Sun, Eta Carina, which is
located in our galaxy. Located in the southern hemisphere, it has been very
unstable for the past 300 years. An outburst in 1843 made it the second
brightest star in the sky (magnitude -1), but it is now only barely visible
to the unaided eye (magnitude +6). ‘It is likely to be the next galactic
supernova,’ says Jeff Hester, a member of the team working on the wide-field/planetary
camera at Caltech. He adds that it will probably resemble the Crab supernova.

Nearly 7000 stars occupy an area 0.8 light years across in the centre
of the globular cluster M15, says Tod Lauer, a scientist at the National
Optical Astronomy Observatories in Tuscon, Arizona, who analysed an image
taken by the wide-field/planetary camera. Some astronomers had expected
to find a black hole at the centre of the cluster, which ground-based telescopes
could not resolve into stars. However the uniform distribution of stars-about
a million times higher than the average density in our galaxy-showed no
sign of a black hole.