ÐÓ°ÉÔ­´´s at the Los Alamos National Laboratory in New Mexico have
developed software which converts a satellite image containing billions
of pieces of information into a representation of only a few hundred numbers
that can be easily manipulated and analysed by a personal computer.

In Britain, a team from Bradford and Reading universities, Cranfield
Institute of Technology and the Natural Resources Institute (NRI), the research
arm of the Overseas Development Administration, have developed a robust
and cheap satellite receiver system which can be run by a personal computer
rather than the usual mainframe. The system, which will shortly begin trials
in Africa, will cost around one-third of the $60 000 or more required for
a conventional receiver system.

The receiver is designed to use data from satellites operated by the
US National Oceanic and Atmospheric Administration (NOAA) because its images
are good for monitoring natural resources and vegetation and, most importantly
for developing countries, they are free.

The system’s low-tech exterior looks more like an industrial heat extractor
than the normal satellite dish. The horn focuses the signal with glancing
reflections to a receiver at the end, unlike a normal parabolic reflector
which concentrates the signal at its focus.

The horn is steered by hand to follow the satellite as it moves across
the sky. NOAA satellites take just 15 minutes to pass from horizon to horizon.
A buzzer sounds when the satellite appears and a voltage meter shows the
strength of the signal being received. When the strength passes a certain
value, recording can begin and the horn is steered by watching the meter
and keeping the level as high as possible.

Most applications in developing countries require a detailed understanding
of a relatively small area. ‘This means that very small amounts of the data
stream need to be received and processed,’ says Jim Williams of NRI. Of
the 100 megabytes transmitted down during one pass of the satellite, as
little as 1 per cent may be used. So the team developed software which can
select the data of interest.

By going back to basics, Williams believes he and his collaborators
have developed a technique which allows freely available data to flow where
it is most needed.

The American system processes images from the American Landsat satellites.
But each satellite image, covering roughly 100 by 100 nautical miles, contains
48 million picture elements, or pixels. Each pixel can be one of millions
of possible colours.

The technique uses a process called ‘clustering’, in which a computer
defines 256 clusters of colours, and assigns each pixel to the cluster it
most resembles. This greatly reduces the amount of data, making it more
manageable.

First the sytem picks 20 per cent of the pixels from the image to analyse.
Then it must find the best match of each selected pixel to the 256 cluster
values. The algorithm uses a decision tree structure where each pixel is
compared to a certain cluster value and it simply decides whether the pixel
is greater or less than that value – something a computer can do very fast.
It only takes 10 such decisions to reach the bottom of the tree, and the
computer then only has to decide whether the pixel belongs to one of two
cluster values.

‘Ours is 100 times faster than previous systems,’ says James White of
Los Alamos, ‘and the results are so good that when we show experts two images,
one of the original data and one that has been clustered, they can almost
never tell the difference.’

Using the new algorithm, White says that a Macintosh personal computer
would take 8 to 10 hours to process a Landsat image.

This in itself is profoundly worrying. The long-term consequences have
major implications for both the industry and all who have worked for it.
But perhaps equally worrying is the damage this will do to the public’s
already fragile trust of scientists and their proclamations. For years the
relevant bodies have refused to contemplate any direct link between exposure
to low-level radiation and cancer. Showing there might be a connection cannot
but prompt allegations that scientists have been less than open, at the
same time lending credence to tomorrow’s horror stories, the evils of genetic
manipulation and the excesses of embryonic experimentation. Any public sympathy
the nuclear industry may have gained when news of its untimely castration
broke late last year will have been lost.

To be sure, if the public had read the small print over the years, it
would know that scientists individually have never ruled out the link between
radiation exposure and cancer. The question has simply concerned the threshold
and the time or intensity of exposure. However, this will be cold comfort
to those affected. Persistent denials from the authorities over the years
will have a peculiarly hollow ring.

The nuclear industry still has to shake off fully its legendary cloak
of secrecy. This first appeared in the 1950s, when years elapsed before
incidents were reported, and although glasnost may have reached the nuclear
industry in Britain and the US sooner than in the Soviet Union, much work
remains to be done to remove the bitter aftertaste.

A feeling of a lack of openness occurs in other areas too. Veterans
of atom bomb tests in Australia and victims of asbestosis around the world
have had to wait years, if not decades, to begin compensation claims. All
the while the scientific community’s integrity loses out as its research
is cited by institutions which many would say were trying shirk their responsibilities.

Now I don’t believe the nuclear industry is inherently dishonest or
genuinely trying to hoodwink the public. Certainly since the publication
of the Black report in 1984, it has opted a more visibly open policy. But
perhaps to regain the public’s trust it needs to do more and actually initiate
action. At stake is not just the future of nuclear power. To squander the
public’s goodwill as we enter an age when all major issues with social or
ethical consequences will have significant scientific components is surely
shortsighted.

How can this change come about? First, it would help if the industry
or company involved in contentious fields did not always seem to be waiting
for absolute proof of a direct link between cause and effect. Science does
not deal in absolutes, especially in the areas of risk and probability.
It is like copying the tobacco lobby’s line that there was no direct link
between smoking and lung cancer.

It was a good sign that it was the National Radiological Protection
Board, rather than a pressure group, that called for a halving of the ‘safe’
level of radon in homes in January. In the same way institutions involved
in contentious fields could lead the way in calling for more research, and
providing funds. It could be that this is going on already, but it is not
what the public perceives.

The current enthusiasm for green issues has ushered in a climate in
which companies are falling over themselves in an effort to prove the colour
of their commitment. If it makes business sense to court public opinion
using green tactics, how much more appropriate would it be for the scientific
community to make a public show of openness and facing up to difficult decisions?
I believe that the Perrier company’s reputation for openness will have been
enhanced by the prompt withdrawal of all bottles worldwide in the wake of
the recent benzene scare. Maybe scientists could see the same happening
to their reputation for objectivity if they persuaded their paymasters that
the future lies in free communication and subsequent action, rather than
muted inaction.

Susanna Lithiby is a freelance writer.