The Crowd and the Cosmos: Adventures in the Zooniverse
Page 15
Brooke on bulgeless galaxies—are just two examples of the
things that are made possible by the careful classifications pro-
vided by Galaxy Zoo volunteers. It’s been incredibly satisfying to
watch the project team use our volunteers’ efforts to understand
more about the Universe, and to see the use to which other
people have put them.
I have, though, continually been distracted while the frenzy of
astrophysical research unleashed by the availability of the Galaxy
Zoo results was playing out. Almost as soon as the project started
116 Into the ZoonIverse
I began to get phone calls and emails from other scientists, in
fields about as far removed from my own as is possible to imagine,
who wondered whether our Galaxy Zoo volunteers might be
willing to help them out, too. The enquiries ranged from the
polite to the pleading, but they revealed what I should have
known already. Astronomers were not the only researchers
struggling with the sheer volume of data now accessible to them;
whether ancient historian or zoologist, they were likely suffering
from the same set of problems, and citizen science, Galaxy Zoo-
style, seemed like a way out.
By this point—somewhere in 2008, a year after the launch of
Galaxy Zoo—I’d abandoned any pretence of doing the work
Oxford had employed me for in the first place, and was thinking
about citizen science full-time. We were given a small amount of
money by Microsoft, part of a fund set up by the company to
commemorate their leading computer scientist, Jim Gray, who
had vanished while sailing near San Francisco in early 2007. (Our
project appealed partly because of the response to Gray’s disap-
pearance; his colleagues and friends organized a distributed
search for signs of his boat in satellite images, a task which was
eerily reminiscent of the kind of thing we’d ask Galaxy Zoo vol-
unteers to do in the years ahead.) A grant from the wonderful
Leverhulme Trust followed (I love an application form which
includes the question ‘why won’t anyone else fund this?’), and for
the first time we could think about expanding.
From this point on ‘we’ includes a diverse cast of wonderful,
slightly bonkers web developers and escaping scientists who
deserve a lot of credit for the last ten years. This isn’t a formal history, so I won’t stop along the way to describe who did what, but
you should be very aware that this is a team effort. My first hire
was Arfon Smith, a cheerful Welsh presence who I’d first encoun-
tered when we were PhD students. Arfon studied astrochemistry
Into the ZoonIverse 117
from a chemist’s perspective, but had realized research was not
for him and become a web developer. (He now heads the grandly
titled Data Science Mission Office at the Space Telescope Science
Institute in Baltimore, the home of Hubble, and so his attempt to escape academic research hasn’t gone brilliantly.)
With Arfon’s expertise on board we could build something a
bit less jury-rigged to support Galaxy Zoo, and also try out some
new things. We were planning the Zooniverse—a platform con-
sisting of many projects—rather than just a single website. As
we thought hard about how to build a system that could be what
we wanted, we slowly acquired a set of test projects, all ambi-
tiously different from each other. One of the first came when I
found out that the team at the Royal Observatory Greenwich
were thinking of developing a similar project using solar data.
I’ve always loved visiting Greenwich—arriving there on the boat
which runs from central London and spying the green dome of
the old observatory up on the hill, behind the beautiful old
Royal Naval College and Queen’s House, is absolutely thrilling.
I was also intrigued by the idea of working within a museum;
the success of Galaxy Zoo had meant we were suddenly com-
municating with a huge number of people in a fairly novel way,
and the idea of a place where there were experts in communi-
cating with the public seemed useful. More to the point, as far as
I was concerned if anyone was going to do such a project it was
going to be us.
I recruited Chris Scott (then at the Rutherford Appleton Lab,
now at the University of Reading), who I’d interviewed several
times on the topic of solar weather. Chris is one of the team
behind a very special pair of cameras, the Heliospheric Imagers
(HIs) on the twin STEREO spacecraft. STEREO’s mission was to study solar weather, the activity on the surface of our star which
can affect the whole Solar System. At any given time, particles
118 Into the ZoonIverse
are flowing away from the Sun in a stream known as the solar
wind, but occasionally things get more spectacular.
The Sun is a ball of ionized gas (otherwise known as plasma).
That means that as it rotates, it doesn’t do so in the way a solid
body does. If you could stand at the solar equator (a terrible idea
for many reasons), you would complete one rotation every 24.5
Earth days. If you stood at the solar poles—a no less terrible
idea—it would take thirty-eight days to rotate. The Sun also has
a strong magnetic field, and this differential rotation has a pro-
found effect on it. The magnetic field becomes tangled, and every
so often releases energy by springing back to an untangled form,
expelling material into space as it does so.*
These events are known as coronal mass ejections, or CMEs
for short (Plate 7). In Chris’ phrasing, each consists of a billion
tons of matter moving at about a million miles an hour. They are
spectacular and dramatic, but they are interesting for practical
reasons too. Every so often, the Earth happens to get in the way
of one of these CMEs. If conditions are right (again, the details
depend on the complexities of interacting magnetic fields and
charged particles), the particles from the CME can cause a change
in the Earth’s upper atmosphere, creating glorious displays of
what are called the aurorae, the Northern and Southern Lights.
The background flow of particles from the Sun means that at
least a faint display of aurora is visible on most nights. I used to
act as a tour guide on special flights to go view the Northern
* The actual physics of this are, from my perspective, unbelievably complicated, and as a result I have the utmost respect for the scientists brave enough to take on trying to understand the Sun as their life’s work. If you want to bamboo-zle most astronomers, just ask if they have considered the effect of magnetic fields; the answer is almost always ‘no’. I have the liberty in my work of looking at distant stars and galaxies and deciding that they look simple; with the Sun, we have no such option and must confront its complexities.
Into the ZoonIverse 119
Lights; we’d fly north towards Iceland, turn the lights in the plane
off (and sometimes those on the plane’s wings too), and peer out
of the windows. I took something like forty flights, and only once
did we have a complete failure. Most of the time, though, what
we could see was a
faint, grey curtain. It might flicker a little, and change shape over the course of the hour or so we’d watch, but
my job as the on-board astronomer was to make sure people
were excited by seeing something so unspectacular.
After all, most of the people on board the flights were there
because they’d seen footage or photos of brilliant and brightly
coloured aurorae, lighting up a snow-covered landscape with red
and green shadows as the lights dance overhead. I’ve seen such a
display only once, on a trip to Tromsø in northern Norway. Local
aurora expert Kjetil Skogli had taken our group out to a frozen
lake, but as we headed to the site we could already see something
spectacular was happening.
Before too long, I was lying on my back in the snow looking up
into a clear night sky that was like nothing I’d ever seen before.
The horizon was lit up with a bright green curtain that seemed to
change ethereally even while we looked at it. Bright streamers
reached up, high into the sky, suddenly brightening and fading as
I looked. Eventually, the sky far above me was encircled with red,
a feature known as an auroral crown (Plate 8) which I’d only read
about before. It was utterly magical, a transformative hour or so
that I will never forget, a few moments’ glory powered by the
arrival of a CME, with particles from one of these events pouring
down onto the Earth’s atmosphere and exciting the particles
there to glow brightly for our entertainment.
The next night we went back out, and drove through a blizzard
to the Finnish border to find a gap in the clouds. Despite this, we
were rewarded only with a faint glow. The Sun, and its interaction
with the Earth, is capricious in the extreme. Yet particularly large
120 Into the ZoonIverse
or energetic CMEs can have consequences that reach beyond the
success of a sightseeing trip. The electric currents induced by
such activity are a serious threat to much of our electronic infra-
structure; a power blackout that affected large parts of North
America and Canada in 1989 was blamed on a solar storm. The
famous Carrington Event, a dramatic flare observed during the
nineteenth century, affected telegraph systems—the high-tech
communications infrastructure of the time. Much of our infra-
structure is now in space, with satellites in the firing line and vulnerable to the effect of CMEs.
With warning, most of these negative consequences can be
prevented. Understanding solar weather, and predicting whether
an observed event might hit the Earth, has thus become a prior-
ity. To this end, the STEREO mission was supposed to provide a unique perspective. It consisted of two separate spacecraft, each
placed on an orbit which meant it drifted slowly away from the
Earth. One was ahead of our planet and one behind, flying with
cameras turned to study the Sun and its environment.
The HI cameras the twin spacecraft carried had a different job.
They were designed with a series of internal baffles, made of
some of the blackest material available, all in the service of reduc-
ing internal reflections. That’s necessary, because these cameras
had the job of staring at the space between the Sun and the Earth,
watching for the faint trace of coronal mass ejections travelling
through space.
The images from the HI, turned into low-resolution video, are
strangely beautiful. You see nothing but a background starfield
at first, drifting slowly past the camera as the spacecraft moves.
You might notice a couple of stars that are much more brilliant
than the others. These aren’t stars, but planets—Venus, or even
the Earth drifting slowly through the field of view. The fact that
we can launch a spacecraft capable of capturing a beautiful
Into the ZoonIverse 121
image of our own planet, as just one drifting object among a
myriad stars, is something that stops me dead in my tracks from
time to time, but this—and the occasional spectacular movie of
a comet having its tail removed by the fast-moving particles of
the solar wind—is very much beside the point. The main goal for
the STEREO HI imagers, and the other cameras on board, was to understand how the solar wind gets launched and then travels
through the Solar System.
Picking out the ghostly trace of a passing coronal mass ejection
against a background of stars is difficult—exactly the sort of pat-
tern recognition task that computers still struggle with and at
which our biological, evolution-honed senses excel at. With the
help of those in Greenwich, Chris’ team, and others, we created a
project, Solar Stormwatch,* which asked volunteers to watch vid-
eos, spot CMEs, and trace their progress across the Solar System.
This was a difficult task for us, as we had to build a whole new
set of tools capable of dealing with video and allowing the care-
ful marking that was needed to produce scientifically useful
results from such a task. It was also difficult for the volunteers,
who had to look carefully to find even the faintest traces of activ-
ity in very busy images. Of special importance were the first few
frames of any particular event, when the particles that made up
the CME had just been launched from near the surface of the
Sun; as they travelled through the lower atmosphere of their star,
they would have interacted with the magnetic fields that thread
the region, producing what could sometimes be a dramatic
effect.
* Most of the development was done by the talented Jim O’Donnell, then a web developer in the team at the museum in Greenwich but more recently a stalwart of the Zooniverse team in Oxford.
122 Into the ZoonIverse
To our delight, Solar Stormwatch was a hit. While not receiv-
ing the publicity that Galaxy Zoo had, tens of thousands of
people took part, and there were a couple of immediate scientific
results from the Solar Stormwatch project. First of all, Chris’
team showed that—partly because of the unique vantage point
afforded by the two STEREO spacecraft and partly because of the
care taken by volunteers to achieve incredible accuracy—using
classifications from citizen scientists provided a better warning
of the approach of a CME towards Earth than existing automated
systems. If you’re a company with commercial satellites, you
would literally be better off consulting our crowd (and, yes,
launching your own version of STEREO, as the vagaries of orbital mechanics have since seen the two spacecraft drift to less useful
positions) than relying on your resident machine-learning
experts. Machine learning for this sort of problem is still in its
infancy, and professional forecasters, it turns out to my surprise,
still inspect the data by eye, but perhaps the Solar Stormwatch
results might provide a gold standard set on which future storm-
hunting robots could be trained.
There was also a scientifically interesting result. I described
CMEs earlier as if they were the result of a sudden event, after
which they just coast out into the Solar Sys
tem. Instead, our data
confirms what had been seen in images from the SOHO satellite further from the Sun’s surface; the particles that make up the
CME accelerate away from the surface of the Sun. Rather than
just setting off into the Solar System, they get pushed on their
way by the complex magnetic fields that exist close to the solar
surface.
Or at least that’s what seems to be going on. Unfortunately,
I for one didn’t realize this was going to be the interesting bit, and so we designed the Solar Stormwatch interface without paying
special attention to the few frames the STEREO cameras capture
Into the ZoonIverse 123
around the point where the CME is just beginning to head out
into the Solar System. We’ve fixed that now, and a newer version
of the project has asked classifiers to pay particular attention to
this most interesting of regions, but the results are still awaited
eagerly.
Solar Stormwatch seems a large jump from Galaxy Zoo, but
we were looking in other directions too. I knew that a large num-
ber of planetary scientists spent their days counting craters, the
better to understand the history of the worlds they were study-
ing, and the task seemed ripe for citizen scientists to sink their
teeth into.
The principle behind crater counting is simple. Imagine that,
at some point in the Solar System’s five-billion-year history, the
great volcanoes of the Martian range sputtered into life, with lava
flowing out to cover the surrounding plains. The new surface
would be, at least when seen from orbit, smooth and new. Wait a
few million years, though, and a meteorite large enough to sur-
vive a fall through the thin Martian atmosphere is likely to hit,
burying itself into the surface or vaporizing near impact, in either
case leaving behind an impact crater. Wait a little longer, a sec-
ond meteorite might hit. And then another. And another. Even
seemingly rare events become common over the billions of years
of cosmic history.
From today’s perspective, we can count the craters to work
out how old the surface is, at least relative to others on the
same planet. The cratered surface of the Martian highlands, for
example, is clearly older than the volcanic and smooth slopes of
the Tharsis Montes. Play this game on the Moon, and there’s an