Florida
LATE last August, an early-morning phone call roused NASA engineer Roger
Avant from his bed. It wasn鈥檛 good news. Three days earlier at Vandenburg Air
Force Base, Avant had watched as a rocket blasted off into the California night.
On board was his baby鈥攁 $65 million satellite named Lewis, packed
with innovative technologies. As project manager for Lewis, Avant had finally
begun to relax.
But now his flight operations manager was on the phone, telling him that
something was dreadfully wrong with Lewis. The spacecraft was spinning slowly
but dangerously out of control, its solar array out of alignment with the Sun.
By the time technicians had spotted the problem, Lewis鈥檚 emergency battery had
run dry.
Advertisement
Over the next few weeks, NASA and TRW, the Ohio-based aerospace
company that built Lewis, worked in vain to cram life-saving software commands
into the satellite during those moments when its solar array was facing the Sun.
Finally, on 28 September last year, Lewis burnt up in the atmosphere.
鈥淚t was a very, very bitter disappointment,鈥� recalls Avant. NASA and TRW are
still investigating what sent the satellite out of control.
When Lewis died, it carried with it the very latest in Earth imaging cameras.
This mission was the first deployment of a sensitive, all-seeing camera called
the Hyperspectral Imager (HSI), designed to survey the Earth鈥檚 surface in
unparalleled detail across a broad spectrum of wavelengths. Had Lewis survived,
design engineers might have learnt whether they had used the correct technical
approach to hyperspectral imaging. Now NASA has to decide when the instruments
that were destroyed with Lewis will get another chance to show their paces. If
all goes well, a similar imager could be in orbit by the middle of next
year.
Nowhere to hide
Unlike a standard camera, the HSI records a large number of images for each
section of the Earth that it surveys. Each one of these images shows only a
narrow band of colour about 5 nanometres wide. These thin slices of
colour鈥�384 of them in all鈥攃over the range from the blue end of the
visible spectrum at 400 nanometres, through red, and into the infrared region as
far as 2500 nanometres. No photographic film would be capable of recording such
a wide range of colours, so the HSI is fitted with sensitive electronic light
detectors, similar to the light-sensing chips in video cameras.
The images at these different wavelengths can then be combined to produce a
single, three-dimensional colour map. Environmentalists should be able to use
these maps to monitor pollutants leaking from a chemicals factory, or to keep a
check on the health of jungles or fish stocks. The data might even allow
biologists to distinguish between different plant species in different areas.
Mining companies could use the data to search for the telltale spectral
signatures of rare minerals. Earth scientists might deduce the age of ice at the
polar caps. And the military has a strong interest as well: tanks and missiles
should stand out against their surroundings like beacons.
All in all, the HSI should reveal far more detail of the Earth鈥檚 surface than
has ever been shown from space before. 杏吧原创s might even detect animals in a
forest, not by their shape, but by the particular colours or spectrum that they
reflect back into space. This spectrum would stand out plainly compared to the
light reflected upwards from the surrounding forest.
The NASA designers charged with building the HSI faced a series of formidable
technical challenges: somehow they had to devise an instrument that would
collect light reflected from the Earth鈥檚 surface, split it into its constituent
wavelength bands, and then record a separate image for each one of these bands.
In the end, they managed to squeeze the HSI into a 40-kilogram package about the
size of a personal computer.
The first problem was to find a way to separate the reflected light into its
constituent colours. The simplest route would be to use coloured filters. But
splitting light into 384 bands would require a huge number of filters, which
would make the device complex, bulky and inefficient. Instead, they opted for a
single, mirror-like device called a reflection grating, which behaves like a
cross between a prism and a mirror. Shine a beam of white light onto it and the
light it reflects is split into a rainbow of colours. Robert Green, a
spectroscopist at NASA鈥檚 Jet Propulsion Laboratory in California, believes that
the reflection grating is the only way to accurately produce the required 384
bands.
The secret of the grating is in the thousands of fine, precisely ruled lines
that cover its surface. When white light hits the grating, different wavelengths
are diffracted at different angles, producing multicoloured reflections. You can
see something similar in the iridescent effect produced when light bounces off
the shiny surface of a CD.
But the real trick was to line up each of the coloured bands with 256
photodetectors. For the 384 bands, this makes a total of 98 304 detectors, and
they are packed into an array only 1 centimetre square. To separate the
diverging bands of light cleanly, the distances between the grating and the
individual photodetectors must be calculated to high precision. If the distance
is wrong, says Avant, or if the instrument is jarred during launch, the separate
colours could overlap at the photodetectors. 鈥淵ou wouldn鈥檛 have the spectral
resolution you鈥檙e after,鈥� he says.
Building the tiny array of sensitive photodetectors was a tough task. As one
scientist involved in the Lewis project points out, it took NASA five years to
build the Landsat 7 spacecraft, and that records just seven broad wavelength
bands. In the HSI, all the photodetectors in the array have to react to light in
exactly the same way. Yet the measured spectrum is so wide that there is no
single type of photodetector that can work for all the wavelengths. So while
silicon photodiodes measure the visible and near infrared wavelengths between
400 nanometres and 1000 nanometres, a different type, constructed from mercury
cadmium telluride, is needed for the infrared portion from 1000 nanometres to
2500 nanometres. Each of these photodetectors converts photons into electrons,
which are sent to the satellite鈥檚 computer through an array of tiny electronic
circuits.
Overheated
These circuits pose their own problems. When they are working, says Avant,
they generate enough heat to disturb the HSI鈥檚 sensitive photodetectors. If the
detectors, especially those at the infrared end of the instrument鈥檚 range, get
too warm they generate electrical noise that begins to blot out the weak signals
they are supposed to be picking up. Spectroscopists estimated that the detectors
needed to be cooled to 80 kelvin. Raising the temperature by just a few degrees
could begin to cause problems, so the array was completely isolated from the
satellite鈥檚 electronics.
NASA scientists have already produced hyperspectral images using an
instrument on a converted U-2 spy plane, but Lewis was to have been the first
spacecraft to do so. Airborne imagers can cover this spectral range without
having to be cooled, because even the highest-flying planes operate twenty times
closer to the ground than a satellite, so the reflected light is much brighter.
鈥淚n space, the amount of light reaching the collecting aperture is far less,鈥�
Avant says, so the readings are much more susceptible to interference. Without
cooling, the signals from the spaceborne instrument will be swamped by
background noise and the data would be useless.
But cooling objects in space is far from simple. The HSI aboard Lewis was
equipped with a special helium cryocooler that had taken 15 years to develop.
The photodetector array was mounted at the tip of the cooler鈥檚 鈥渃old finger鈥�, a
protrusion built to draw heat away from the array, and at the same time keep the
device as far as possible from the spacecraft鈥檚 other equipment. Some of that
equipment, especially the computers, had to be warmed up to operate properly, so
designers thermally isolated the two sections of the spacecraft.
Many coolers operate by circulating cold liquid through a network of tubes,
but engineers working on Lewis quickly ruled out that strategy. 鈥淚f you had
flowing fluid, the friction with the walls of the tubes would be acting as a
heat source,鈥� Avant says. The challenge was to design a cooling system that
would all but eliminate internal friction.
So engineers decided to generate a cooling effect by compressing and
expanding helium gas. Lewis鈥檚 pulse tube cooler, a device similar to a pair of
engine pistons, produces these compression waves mechanically. In regions of low
pressure, where the coolant expands, the temperature drops in exactly the same
way that gas from a compressed-air cylinder or a bicycle tyre cools as it
escapes into the open. Creating this expansion in the cold finger draws heat
away from the photodetectors and cools them. The heat could then be radiated
harmlessly into space.
But this strategy added a major complication: how to introduce compression
waves without vibrating the detector and ruining the readings? Excessive
vibrations would also damage the cryocooler, possibly allowing vital helium to
leak away.
Data avalanche
The solution, the engineers decided, was to fight vibration with vibration.
They added an electronic device to the cooler which could precisely measure the
vibrations imparted by the compression waves. They then introduced equal and
opposite vibrations to cancel out the movement, an idea borrowed from commercial
manufacturers.
鈥淭hese sort of techniques have been used in factories for many years.
However, their control systems were usually large and power-hungry,鈥� Avant says.
The trick was to squeeze the electronics into a satellite.
Coming up with a system to manage the huge quantity of data flowing from the
HSI was another big challenge. The imager works like a broom with a single row
of 256 bristles, says Green, each bristle representing a patch of the Earth鈥檚
surface 30 metres square. Behind each bristle sit 384 detectors to record the
complete spectrum from each of these squares. Sweeping across the surface, the
imager would record a succession of transverse strips almost 7.7 kilometres long
and generate 440 megabits of information each second.
If the imager were designed to be carried aboard a relatively slow-moving
aeroplane, this torrent of data might be manageable. But Lewis鈥檚 orbit should
have kept it whizzing about the Earth鈥檚 poles at high speed, imaging an area of
almost 60 kilometres square every second. 鈥淭hat leaves only a few hundredths of
a second to read out the data from the array and get ready for the next
snapshot,鈥� Avant says. 鈥淭he big challenge is the rate at which the data needs to
be stored.鈥�
However, TRW and NASA did manage to perfect a flight computer and data
recorder that could do the job. As the lines of data were read out from the
spectrometer, they were packed into a solid-state, digital recorder capable of
holding 4 gigabits of data in its memory. Lewis was supposed to beam the data
down to eager scientists each time it came within range of TRW鈥檚 ground station
at Chantilly in the suburbs of Washington DC. 鈥淭he flight recorders of the past
simply could not carry enough data,鈥� Avant says.
For many members of the Lewis project, the real challenges would have begun
once the data reached the ground. 鈥淏ecause you鈥檙e looking at so much
information, how do you sort through it and find the nuggets?鈥� says Granville
Paules, head of advanced planning at NASA headquarters in Washington DC.
The information would also have had to have been calibrated to correct for
the distorting effects of the atmosphere on the light reflected from the Earth鈥檚
surface. The atmosphere refracts different wavelengths of light by different
amounts. So scientists estimated how each wavelength would be affected as it
travels up through the atmosphere, and equations or algorithms were then
developed for each of the 98 304 photodetectors to compensate for these
effects.
To help improve the images further, Lewis was equipped with a black-and-white
camera that could record objects as small as 5 metres across. The idea was to
use these detailed pictures to align and sharpen the lower-resolution,
multicolour images from the HSI. Green helped to develop the algorithms to
calibrate these images from the satellite.
The researchers have also developed software to search the data for unusual
or distinctive wavelengths that might indicate valuable features on the ground.
Paules calls this 鈥渟ub-pixel identification鈥�. Although each pixel corresponds to
a relatively large 30 metre by 30 metre patch of ground, the presence of an
animal or a rare mineral would still change the spectral signature of that
pixel. 鈥淵ou鈥檙e checking to see if there鈥檚 a light signature in any one of those
384 bands,鈥� Paules says.
The images could hint at features that are much smaller than 30 metres
across. 杏吧原创s at NASA鈥檚 Jet Propulsion Laboratory in Pasadena, California,
have developed software that sorts through the images for important background
signals. Mineral prospectors or defence analysts could then deploy other sensors
aboard satellites or aircraft to take a closer look.
Sadly, the premature destruction of Lewis means that the HSI has yet to show
whether it can live up to its creators鈥� expectations. But that may soon change.
NASA has called a meeting at its Goddard Space Flight Centre to decide which of
the technologies aboard Lewis should be included in the next hyperspectral
imaging project, called EO-1, which is scheduled for launch in May next year. If
the HSI gets the go-ahead, Avant will soon be in for a few more sleepless
nights.
