Horizon (1964–…): Season 45, Episode 16 - Why Can't We Predict Earthquakes? - full transcript

It's Sunday 11th May, 2008.

For millions of people,
it's just another ordinary day.

Meanwhile, at earthquake monitoring
stations around the world,

there's nothing special to report.

What no-one knows
is that 24 hours later,

an extraordinary natural disaster
is going to strike.

TV: A massive rescue operation
is under way...

..7.28 this morning,
sending shock waves around Asia.

..a magnitude of 7.8
struck Sichuan Province...

An enormous earthquake tears into
Sichuan Province in Western China.

Over 50 million people
are affected.



Five million lose their homes
and 70,000 die.

And all because science can't answer
what seems like a simple question.

In the last 100 years, earthquakes
have killed over a million people.

And with the growth
in the world's population

scientists predict that this century
might see ten times as many deaths.

So why can't we work out
when and where the next big quake
is going to happen?

Well, the more work we do
on earthquake prediction

the more difficult
it seems it's going to be.

You start to think you see patterns
and understand them

and then when you try to
play the game forward

and look for those patterns,
it just hasn't ever panned out.

If you were a seismologist and
you knew how to predict earthquakes,

er, you've arrived.

So why is earthquake prediction
so difficult?



And what is science doing
to overcome this force of nature?

If you want to know
about earthquakes,
this is the place to come -

California.

# They tell me the faultline

# Runs right through here... #

American's golden state
lives in the constant shadow
of an enormous earthquake,

and because of this
they throw more money into studying
these disasters than anywhere else.

# They tell me the faultline

# Runs right through here. #

At the heart of this effort is
the United States Geological Survey,

the nerve centre of
earthquake monitoring.

This is a map showing the global
earthquakes of the last week.

Any time you look at this picture
there's going to be aspects of it
that are gonna look a lot like this.

You're gonna see this distribution
down California

because this map is showing smaller
magnitude earthquakes in California.

The red shows that we've just
had an earthquake, this is at
the Northern end of Japan.

There's usually something
in Japan every week that's
large enough to show up here.

The South Pacific is one of the
active areas of the world right now.

You'll see a few things spread
around, somewhere through Asia
essentially all the time.

For seismologists like Lucy Jones,

it's no longer a mystery why
the earth suffers so many quakes.

The tectonic plates
that make up the world's crust

grind against each other,
building up huge amounts of stress.

The stress produces cracks
known as faults.

Wherever there's a fault, there
might one day be an earthquake.

We know that earthquakes happen

because stress
builds up in the crust

and finally you overcome
the friction and you slip suddenly.

Actually, an analogy
is snapping your fingers.

When you snap your fingers you have
two surfaces in frictional contact.

But, all right, now I'm trying
to say what micro-second
they're going to move on

and that's going to be exactly
what point the friction is overcome.

That's, er... There's a lot of
processes going on there.

But even though scientists know
how and where earthquakes happen,

the question they can't answer
is the one that matters most.

So on May the 11th did you think
there was going to be
a big earthquake in China?

No. There's nobody who on May 11th
said there's gonna be an earthquake.

There are plenty of people
on May 13th said "I really
did know this two days earlier."

It's a far cry from the picture
of just couple of decades ago.

This is Parkfield,

a tiny hamlet
in the middle of California.

Few people ever come here,
and even fewer stay.

But this village lies on top
of the infamous San Andreas Fault,

and once looked like it held the key
to understanding earthquakes.

It all began when a team of
geologists led by John Langbein

noticed something unusual
about little Parkfield.

It was, er, in the '70s and
early '80s it was recognised

that there was a sequence of
magnitude six earthquakes

that repeated the same stretch of
the San Andreas Fault

every 20-odd years and it didn't
take too much imagination

to extrapolate and say the next one
ought to be in the late '80s.

The village had always suffered
from earthquakes, but these quakes
followed a very distinct pattern.

Langbein's team decided to
use Parkfield for a bold
and unprecedented experiment.

To see what happened to the ground
before an earthquake struck.

Estimating that the quake
would occur between 1987 and 1993,

scores of geologists
descended onto Parkfield.

They came to town and they
set up shop with instruments,

and they're sort of hidden away
and tucked away

so they're not that obvious but
there's a lot of them out there.

And the idea was to have
the instruments ready to catch
the next earthquake red-handed.

Well, what you're hoping to see,
the analogy is a stick breaking,

so in the long term
you're bending the stick,

you see it, er, deform

and then maybe just before the stick
actually goes snap

you'll hear "crack crack crack"
or something like that.

Now that they had narrowed
down the time window

and knew where
it was going to strike,

this was science's best chance yet
to see an earthquake in action.

Millions of dollars flooded in
to fund the research.

All they had to do now
was just sit and wait.

This ranch house is the high-tech
outpost for a team of scientists
from the US Geological Survey.

Every morning they check
sophisticated sensors,

looking for signs that the Earth
is ready to rumble.

So how many instruments
are buried here in Parkfield?

You know,
it's a little hard to count.

There's probably about
two to three hundred.

We had some creep meters
that measure fault slip,

some geo-chemical experiments,

strain metres,
pole positioning system.

TV: And on the fault line itself,
TV cameras are constantly recording.

The instruments
may provide a perfect...

As 1993 approached, excitement
mounted amongst the geologists.

TV: For five years scientists have
been preparing this experiment

for the quake of '93. Now that it's
built, they're hoping it will come.

But then, 1993
passed without incident.

'94.

'95.

'96.

There was still nothing.

Our guess was basically,
what you'd call...

um, ambitious or optimistic.

It wasn't until September 28th 2004

that the earthquake finally struck,

and when it came, it wasn't what
the scientists were hoping for.

It was like the fault
was quiet quiet quiet

and then it broke,

and it was sort of,
it was a fairly negative result.

You know, we were sort of waiting
to catch that precursor

with all these instruments,
and nothing happened.

Instead of finding signals
that might predict an earthquake,

all that the Parkfield experiment
seemed to prove

was that these natural disasters

were far more complex
than anyone had ever imagined.

You know it was sort of taken
as a negative result

and some people were saying
"Time to put the nail in the coffin.

"Earthquake prediction is dead."

And I think that's a bit extreme.

One possibility is that earthquakes
are different.

In '66 there was quite a large
foreshock and in '34
there was quite a large foreshock

and in both of those cases
the quake started up here
and went that way.

And in 2004 there was no foreshock
and it started in the South

and went the other way.
So earthquakes are complicated.

Many people thought that Parkfield
might solve the mystery of
earthquake prediction for good.

But instead it was back to
the drawing board for science.

Holy shit!

Holy shit.

Oh, my God.

Holy cow.

And as every year goes past,
more earthquakes continue
to plague our planet.

Out here, out here! What happened
to the telescope? Destroyed.

Go, go, go, come on!

It may fall.

I got it on tape!

Holy shit.

I got it on tape.

Go, come on.

In the last decade alone,
tens of thousands have died

in Turkey, India, Iran and Pakistan.

And it was an earthquake that caused
the Boxing Day Tsunami,

killing a quarter of a million
people in 2004.

Then in May 2008, it happened again.

This time in Sichuan Province,
China.

Six months after the disaster,
geologist Mike Ellis

is travelling to China
to investigate the earthquake.

Mike has worked in this region
before

so he always knew there could be
a major quake here one day,

but he didn't think he'd ever
get to see it in his lifetime.

I've chased quite a few
earthquakes, as we call it,
um, Taiwan downwards.

Big earthquakes like this happen
in the ocean all the time but of
course you can't go there to see them

so academically and scientifically
it's a treat

to come to an earthquake like this

but of course it's a very
sobering experience as well.

For the inhabitants of Sichuan
Province, Monday May 12th
was a day like any other.

Many people were at work,
their children in school,

while others were simply
out enjoying the sunshine.

Little did they know
that the ground beneath their feet

was about to be ripped apart
by a rupture that would travel

100km in just 50 seconds.

REPORTER: A massive rescue operation
is underway after a powerful..

..thousands are killed
after a massive earthquake
hit South West China.

At magnitude 7.9,

it was one of the world's most
powerful earthquakes in decades.

But even after the shaking
had stopped, the real drama
was only just beginning.

As entire towns collapsed,
thousands of people

were crushed to death
or killed by falling masonry.

And strong aftershocks,
some higher than magnitude six,

continued to strike
across the region
causing new casualties and damage.

REPORTER: The rescue operation
is one of the biggest ever.

50,000 troops
have been mobilised....

..mourning for victims of
the Sichuan earthquake.

Rescue work continues but very few
victims are being found alive.

Months on, and Sichuan
still lies in ruins,

but Mike hopes to find answers
among the broken houses
and upturned soil.

He'll be travelling with Jing Liu,

a Chinese earthquake geologist who's
been mapping the rupture since May.

The county town of Beichuan is 138km
from the earthquake's epicentre,

but it lies in one of the
worst-affected areas.

12,000 people died here,

three quarters of the population.

Coming back for the first time
since the earthquake,

Mike is struck by what's happened
to the place he once knew well.

It's very sobering.

Um, not something that
you want to see, really.

On the river, there were some
trees and a cafe down there,
I used to sit and play Mahjong.

Now it's completely chock full
of sediment and, er...

Mike came here before to map some of
the earthquake faults in this area.

This is one place where I think
the rupture did coincide quite well
with the fault, the mapped fault.

We mapped along there and then
through the valley and up over there.

By mapping faults, Mike hopes
to predict where and even when
an earthquake may strike again.

Now he has his best opportunity
in years.

The recent earthquake has revealed
faults never seen before.

Even though Mike
is hundreds of kilometres
from the tectonic plate boundary

between India and Asia,
this is major earthquake country.

Geological maps of the region
suggest that
there are thousands of faults

hidden in the mountain range
that fringes Sichuan Province -

the Longmen Shan.

So here you can
see the big picture

of the India, Asia collision region
I suppose you could call it.

White area for a high elevation
and the darker areas are lower.

So here is the Himalayan arc,

India of course, moving up into,
into Eurasia,

and that's occurring about
40, 45mm per year,

which is pretty fast
in plate tectonic terms.

There's a series of thrust faults

that come down and around
the Himalayas because

this is the plate boundary

and so there's Longmen Shan

and it's facing the very flat
and relatively low Sichuan basin.

It actually is, geologically
this is a wonderful enigma.

It's always exciting to find a place
that has not been explained yet.

I think we're all looking
for something that we can
make an impact with.

Over in California, most of the
mapping work has already been done.

Here, scientists are all too aware
of the cost of not knowing
where the faults are.

Their wake-up call came in 1906,
when a magnitude 7.9 earthquake

bought San Francisco to its knees.

The violent shaking and fires
afterwards killed thousands

and destroyed much of the city.

Ever since, the state has learned
to live with the threat

that another major earthquake
could strike at any time.

Over 100 years later,
California is now the place to be

for seismologists
and geologists the world over.

There are not a whole lot
of earthquakes in London, you know,

and the ones that happen are piddly
and not worth studying.

So in my game where we measure
things, you'd really like...

something bigger than a seven.

Eight's nice, nine is terrific.

Not so good for people,
but terrific for the scientists.

Roger Bilham was born in England
but moved to America to be nearer

to faults like the San Andreas,
cause of the 1906 earthquake.

So here look, have a look at this,
this beautiful flat valley here

with a hill on each side.
The fault runs right down the middle

and this slipped in 1906.

It slipped only about two metres
here,

as we get further north
it slips increasingly more

until you get
north of San Francisco

where it becomes
about six metres of slip in 1906.

By mapping out California's faults,

scientists are
beginning to understand
how tiny slips in one place

could lead to huge earthquakes
somewhere else.

Roger hopes to pick up a small sign

that might predict
a future disaster.

He travels from fault to fault

checking on a collection
of home-made instruments

that he's buried at sites
up and down the state.

Well, I consider them my babies.

You plant them in the ground and
then they live out their rather
dull but informative existence

sending us information about the
movement of these wonderful faults.

So yes, I quite enjoy it,
except when it rains,
and it doesn't do that much.

Ah, and sometimes the local people
shoot at you, which isn't such fun.

Sorry? Well, yeah.
I work all over the world,

but California is the only place
where they really, really
tried to shoot me,

and, er,
that's the sort of macho people
that go around California with guns.

They sit on these interesting faults
and they don't want you
to measure them.

Astonishing.

We're not going to get shot at
by the owners here, are we?

No, not at all. Sure?

Yes, absolutely,
they're lovely people.

OK, so if you can squint along
this fence you will see it's offset.

Now this fence was put in
after the 1906 earthquake,

so the offset has occurred
since 1906.

And we know from measurements
along the road

and from the creek meter
in the field

that it's moving at about a quarter
of an inch a year, relentlessly,

and when
the San Andreas Fault slips,

this side of the road is going off
to the South, this side of the road
is going off to the North,

so this is stuck next,
this is glued to the Pacific plate

and you're standing
on the North American plate

going away whizzing past me
down sort of Mexico direction.

Let's go and visit the machine.

So this is the important step -

check the bulls
are in the right place.

So it's just over here in the grass.

I sometimes worry
there's a snake under here.

Not this time.

There is an element of, er, a
Heath Robinson contraption about it.

It's a very simple gadget,
it's a cylinder with, um, a rod,

the rod is connected across from
the other side of the fault.

When the fault slips it pulls this
rod away from the metal sensor.

It has a range of about, er...

This one is about 30mm and so
because there's 7mm of slip here,

about every four years
I have to reset it.

But first, download the data.

I'll get my computer out.

7mm of slip
might not sound like much

but it could have
a devastating effect.

150km up the fault in San Francisco
the tectonic plates are locked,

and eventually this pent-up energy

will be released
in the form of an earthquake.

Every time that Roger's machine
measures the fault slipping,

known as a creep event,
this may help to calculate

the amount of stress
that's building up beneath the city.

Oh, we've got a creep event!
How exciting.

The black line here is,
er, can you see that OK?

So the black line
is the temperature decrease

from mid-summer to...
It's upside down, OK,

we could actually turn it up the
other way but let's do it like that,

so there's the temperature
decreasing as a function of time

and here is, er, a creep event

where the fault suddenly starts
slipping at a few
millimetres per second

and then over the next day or two,
in fact continuing
for several weeks.

Even if you'd been standing on the
fault, you wouldn't have noticed it

cos it's really
a very slow, quiet process.

So until people have put instruments
like this on the ground,

we had no idea that these things
were occurring.

Science has mapped
every fault in California,

but in China,
the process is only just beginning.

Xiao Yu Dong is 42km
from the earthquake's epicentre.

What was once flat farmland

was completely transformed
on May 12th.

To find the fault here, Mike's
looking for earthquake scarps -

steps in the landscape where the
rupture has lifted the ground up.

This level I'm standing on right now
used to be up there.

That's a good two to two and a half,
possibly even three metres.

So that entire free surface
is the earthquake scarp.

And no doubt
that will quickly be bricked up

and you won't be able to
distinguish it so easily.

This is also a superb place,
by the way, for finding,

um, unambiguous lateral offset,
if there is any.

You can see where I'm standing,
there's a nice straight wall,

and I can see from here
that the offset,

the natural offset here is about
about a metre,

maybe a little bit
less than a metre to this wall,

so essentially this wall here
was that one back there.

Before the rupture actually happened,

probably at this location there was
a lot of shaking and rolling

and then this side of the
village just rose up like this.

It takes about between 10, 15, 20,
maybe even more seconds to do that.

That's actually quite slow when
you're standing here as an eyewitness

and seeing this thing
just rise up like this

out of the ground and then this
entire part of the village

is now a metre to two metres higher.

It's hard to imagine what
this place once looked like,

but one villager
has kept a memento.

TRANSLATION:
I really liked the beautiful view
we had of the landscape here,

so I took this photograph
from the first floor of our house.

Before the earthquake the road
used to be completely level.

The ground too,
everywhere was level,

but now it's dropped
by one or two metres.

The ground just slid down,
it was amazing.

For the people of Xiao Yu Dong,
May 2008 was the first time

any of them had ever experienced
a major earthquake.

But Mike is beginning to suspect

that there have been
other quakes here in the past.

The two things that important here
is that the elevation difference
between where I'm standing right here

and up there is significantly higher
than it was

further along the rupture

and further back
along the rupture that way
we saw the modern earthquake scarp

being very irregular and this
looks almost identical to that,

so this is very suggestive.

So I would, I would love
to be able to walk,

if I could just walk up here
a long way.

To Mike's expert eye,
every rise and dip in the landscape

could be evidence
of a whole history of earthquakes.

So right here I'm standing on
an old scarp that's very gentle now.

And it continues to be
this sort of hammocky topography

in these fields,
so this sort of old scarp

and the much greater relief here
at this point

would tell us that this has been
the place of an earthquake before

and probably several before that too,

so points to the importance of
mapping these things in great detail.

But mapping out faults
isn't always so straightforward.

Mike and Jing have come to a valley
further along the rupture

where the ground rose up
twice as high as in Xiao Yu Dong.

They're hoping that a scarp
this size will tell them

more about earthquakes
that have happened here in the past

and those that are yet to come.

But recent heavy storms
have transformed the landscape.

TRANSLATION:
Hello, was there an earthquake scarp
here before?

TRANSLATION:
Yes, but it was washed away -
our village was too.

Tons of earth fell down,
part of the mountain just collapsed.

It's probably there,

but not where the river
has incised through it

but if you follow it up a little bit
there must be some remnants...

Maybe, maybe.

I would hope so. A six-metre scarp
can't disappear completely.

That is the scarp.

Let's see, you can go up and I'll
find this place for the grass.

Before this September the ground
was like at this level

and this level used to be lined up
over where, um, Mike is standing.

So I'm standing here on
a surface that now is occupied
by Jing down below there.

The ground on my left
was pushed up six metres

and moved to the right
by six metres as well,

so the mountains as a whole
are shortening,

the crust is shortening
and moving sideways

and that sideways motion is small,

but it is an expression
of India colliding into Asia.

This landscape we're in now has been
formed by many, many earthquakes,

hundreds and thousands of
earthquakes.

Despite the damage from the storms,
Mike is beginning to understand

the history of earthquakes
in this particular valley.

But this evidence
is fast disappearing.

Obviously, over time, the subtle
signals for any specific earthquake

disappears very quickly,
and this has virtually disappeared

in less than six months, so as we're
scrabbling around the hillsides,

we see signals every now and then,
but the data is very sparse,
very difficult to put together.

Geologists like Mike hope that
by mapping the past,

they'll come closer
to predicting the future.

But even if
you know where to measure,

every earthquake is complicated
by a significant factor,

how big it's going to be.

We don't know what makes
an earthquake start today
instead of yesterday.

We also don't know what makes it
stop

and that's what controls
the size of the earthquake.

A magnitude three
starts at a point,

you start to slip at a point
and you have a rupture front that

travels out and causes
more of the fault to slip,

and in a magnitude three,
you travel out this far.

In a magnitude five, it travels
out for a kilometre or two.

In a magnitude eight,
it travels out for 500 kilometres,

so when we're trying to predict
what the earthquake will be,
we're saying, it starts here,

but does it stop after one
kilometre, or does it stop
after, um, 100 kilometres?

Predicting the size of
an earthquake is essential.

Millions of quakes happen
all over the world each year,

but the vast majority are
too weak even to be felt.

The real challenge for science
is to work out when one of these

little earthquakes is going to
develop into a major disaster.

You don't want me to predict
every earthquake, there's going
to be 50 in California today.

You want me to predict which of
the 35,000 we record each year

is the one or two large enough
to do some damage, and really,
what we want to do

is really predict just
the one that happens every five or
ten years that does a lot of damage.

'Live, Los Angeles tonight,
battered and bracing for the worst.'

The last earthquake to do
a lot of damage in California

stuck in Northridge, a suburb
of Los Angles, in January 1994.

Measuring magnitude 6.7,
it killed 72 people and caused over

$20 billion in damage,
making it one of the costliest
natural disasters in US history.

'The earth is literally split here.'

'The city wakes up to a nightmare.'

But one man saw it coming.

Professor Vladimir Keilis-Borok,
an 87-year-old

Russian geophysicist
at the University of Los Angeles

has developed a way of predicting
earthquakes, with a surprisingly
high level of accuracy.

Out of 17 earthquakes worldwide

which happened since '92,

we have predicted 12.

'The Earth's fury

'unleashes fire, and flood,
and fear.'

The prediction method doesn't come
from the world of geology, but from
an extraordinary branch of maths...

chaos theory.

Chaos theory seeks to find an
underlying order in some of
nature's most random processes.

Weather systems,

the way birds flock together,
or even the distribution
of leaves on a tree.

There didn't seem to be any order
to earthquakes, but Keilis-Borok

brought together scientists
from multiple disciplines

to study the problem, including
seismologist David Jackson.

Well, the general theory is that
when the earth is in a chaotic state,

there will be some features
that can be recognised.

And typically, those features are in
the smaller earthquakes that occur,
and how much a small earthquake

brings with it,
some immediate follow-on earthquake.

Looking at some of California's
major earthquakes in the past,

the UCLA team thought that
they could see patterns in the
smaller quakes that preceded them.

Today, they look for similar
patterns, chains of small

earthquakes linked by their size
and the time they strike.

If they think they see a new chain
that matches their historical data,

the group then issues
an earthquake alarm.

Sometimes, humans can see
the patterns and we propose

something that seems to us logical in
terms of the way earthquakes behave,
but sometimes, their patterns are

too complicated and the hope is that
computers, using vast amounts
of data,

and, er, combing the data for those
patterns, can out-think us in that
particular way.

But the patterns haven't always
led to accurate predictions.

Nine years after Northridge,
Keilis-Borok's team announced
that a major earthquake

would strike near Palm Springs
by September 5th, 2004.

Once again, the enigmatic Russian
was putting his career on the line.

But this time, nothing happened.

The team's work continues to be
a mixture of success and failure,

but Keilis-Borok is confident
that he can improve his hit rate.

There is no such thing as

100% accuracy, but...

we believe the accuracy can be
increased by factor at least five.

It remains to be seen if
chaos theory and maths are the answer
to earthquake prediction.

In the meantime, science has
been forced to explore other,

sometimes stranger avenues,
to try and solve this problem.

Since time began, people have been
reporting weird goings on

in the days, or hours, before an
earthquake. Sudden upsurges in
migraines...

..mysterious changes in
ground water levels,

but perhaps the most bizarre
phenomenon involves animals.

Guangxi province, South West China.

This farms lies at the centre
of an intriguing experiment
to predict earthquakes...

..using snakes.

TRANSLATION: We call this snake
Dragon,

or Earth Dragon, here in China.

In Chinese culture, we think of
ourselves as children of the dragon,

so there is no need
to be afraid of snakes.

Jiang Weisong, head of the local
earthquake bureau, has a team

monitoring these snakes 24 hours
a day using webcams.

It's thought that snakes may be able
to sense earthquakes in the same way

that they locate their prey.

Using their inner ears to pick up
vibrations in the ground.

If a small earthquake happens within
120 kilometres of this region,

for example, a magnitude five,

then the snakes will come out of
their holes

and crawl along the walls,
trying to escape.

If a major earthquake happens nearer,
then the snakes would

smash themselves against the wall
continuously,

until they killed themselves.

It has a very powerful effect
on them.

We'd like to see this happen
three to five days in advance,

then we'd have time to analyse it
and make an accurate prediction.

Animal predictions aren't
without foundation in China.

They've been attributed to saving
tens of thousands of lives.

At the height of the cultural
revolution, the city of Haicheng was

evacuated after many people reported
seeing animals behaving strangely.

When a magnitude seven earthquake
struck days later, Haicheng was
heralded as the first time

one of these disasters
had been predicted using animals.

But since then, no-one has ever
been able to replicate the results.

As far as I know, I'm the only
person doing research in this area.

Even in China.

I can understand why other scientists
might not recognise my work, but
I think the reason they distrust it

is that they haven't done
the practical experiments themselves.

If we can have more observation
stations, then our predictions would
be more scientific and more accurate.

One flower doesn't make a spring,
but hundreds of flowers
can definitely make spring.

In the hours, or days, before
an earthquake, it's not just
animals that can be affected.

There's another even stranger
phenomenon that can be used
for prediction.

Bright lights that appear
in the sky.

This photograph was taken
in September 1966,

before an earthquake struck
the town of Matsushiro, in Japan.

Many other people have reported
seeing these lights,

but no-one has ever been able
to prove why they might happen.

Today, however, NASA physicist
Friedemann Freund believes he
may have found the answer.

Gary, will you tell us when
you make contact?

Now, it starts.

In 2005, Freund made a peculiar
discovery that if you crush a rock

to almost breaking point, it
produces a tiny electrical current.

Now, we are already driving

something like four nanoamps

through this rock,

the pressure increases
more and more,

the current increases,

now the pressure has already reached
its maximum value and the current

will stay up there, and as long as
the load stays on the rock,

the current will continue to flow,

and that is the simulation for what
we believe to be happening

in the Earth prior to an earthquake,
before they rupture. If you can
imagine

that you have
a cubic kilometre of rock being
stressed or...the currents translate

into thousands, ten thousand,
sometimes hundreds of thousands
of amperes

that could flow out of
a cubic kilometre of rock.

The currents going through the rock
can give rise to other oddities,
including one that Freund believes

may explain the lights
in the sky before an earthquake.

If it were dark here, we would start
seeing little flashes of light
forming along the edges

of these rocks. Maybe in nature,

they are sufficiently strong that
they couldn't become luminous

phenomenon known as earthquake
lights that can happen before

earthquakes, during earthquakes and
also during the aftershock series.

You think there's a connection
between this and...?

Oh, yes, yes,
there's definitely a connection.

Friedemann Freund is fighting
a tide of opposition from
mainstream science,

but he's convinced that
he's right and he's prepared to
put his money where his mouth is.

So far, everything that I've
shown you was essentially done
on a shoe-string budget,

with lots of private money
going in there

and very, very little funding
from any government sources.

Who's been funding it up until now?

Well, I eventually paid most of it
out of pocket, we are still

having a very, very minimal
funding level.

Just out of your own wallet? Yes,
I've spent close to
a million dollars

on funding this research,
because nobody believed me.

That's a hell of a lot of money
just to try and...

Well, because I know that I'm on the
right track, so I will pursue this

and bring it to the end.
Now, people start to listen,
and yes, now they are convinced.

With something like this,

as clear as you can hope
you would get it.

The scientific community may still
be sceptical about Friedemann
Freund's rock experiments,

but his research is now being used
in a commercial application...

QuakeFinder - a device that
measures electromagnetic changes

in the ground
to sense if an earthquake is coming.

It's, er, basically, a computer
system, er, set of electronics

to process the data, a simple hard
drive from a laptop to record it,

a radio link to bring it into, er,
a farmer's house maybe 200,
or 300 feet away, and then we have

a satellite dish that takes the data

and brings it through a satellite
link up to our site here in
Palo Alto.

It's still early days
for QuakeFinder, but it may have
already had a minor breakthrough.

In October 2007, the little white
boxes picked up electromagnetic

signals shortly before an
earthquake struck Alum Rock.

A small community
south of San Francisco.

This is the, er, the actual data
from the, er, Alum Rock earthquake,

if you're interested in that, these
are the days prior to the earthquake

so the, er, magnetic pulsations
that we see are very, very few
and far between, this large one here

is a calibration signal that
we generated ourselves just to make
sure everything was working OK.

About two weeks before the
earthquake, we started to get these

very large pulsations, the next
few days, it got busier and busier

it spread out over more of the day

until finally, right there,
the earthquake hit.

But was there a moment, Tom,
when this data was, you know,

more and more data's coming in
from Alum Rock, were you thinking,
"Crikey, this must be an earthquake?"

I'll be honest with you, no.

Because we're still trying to
discover what the pattern is,

we're not quite sure how many days
it should be there.

This was, we didn't know if it
was a large earthquake or a small
earthquake, all we knew was that

it was only happening
at that one station,
not at any of the other stations.

It's going to take a great deal of
research and a lot more earthquakes

before theories about rocks
or animals are ever proven.

But mainstream science
has practically given up on
funding these kinds of experiments,

and many geologists even question
the value of prediction.

Well, what would you do with it?
Let's imagine I can tell you

there'll be an earthquake
in a hour, what would you do?

You'd get your camera out,
or your tape recorder or something,

if you were in a building, you'd
probably go outside because you

might think it's gonna fall down,
that's not particularly useful,

the building is gonna fall down,
that is the problem.

Would you rather have an hour to get
out of a building or a building that
didn't fall down in the first place?

It's a real possibility that we'd
have more people dying on the freeway
trying to get away when we made

a prediction than we would have
killed in the earthquake
when it happened.

Unable to predict these disasters,
California has turned itself into

one of the most earthquake-proof
places on the planet.

In Los Angeles,
every new skyscraper has been built
following strict construction codes.

Hundreds of freeway flyovers have
been retro-fitted and re-enforced,
and as the city expands

into the surrounding counties, the
fault lines are what matters when
it comes to choosing real estate.

Would you live directly on it, Ken?

No, when we were looking for
a house, we looked at some houses
that were right on top of

the Sierra Madre fault
and we decided to keep looking.

Similarly, when we were looking
at the house, I was probably

more interested in the structural
integrity of it and the construction
of it than most people would be.

Geologist Ken Hudnut works for
the US GS, preparing Los Angeles
for the next big earthquake.

You can see here
a brand new development going
right up to the Cucamonga Falls.

We think that that fault is
capable of a magnitude 7.5,

7.6 earthquake on its own,

without any involvement of
the San Andreas Fault itself.

That gap is there because they have
to set back away from the fault,

that's the case for any fault
that's considered active,

and by that, the state law says
if it has had

surface faulting within the last
10,000 years, you need to set back
from it.

Over in China, the devastation in
Sichuan Province serves as
a stark reminder

of the potential cost
of building on earthquake faults.

Mike and Jing have come to Bailu,

a mountain town around 50 kilometres
from the earthquake's epicentre.

The fault passes right through
this valley, heading straight for
the town's middle school.

Well, this is quite something.

Thanks to an astonishing
stroke of luck,

the rupture missed both of the
buildings containing classrooms,

but at the end of the playground,

the earthquake demolished a block of
housing, killing several teachers.

What we would, er,
be very happy about seeing here,

um, extraordinarily happy, is that
these buildings that are built either
side of the rupture didn't collapse,

and that one over there
appears to have very little damage,
you know, apart from broken windows,

but, er, this rupture goes through

what used to be the dormitory
for the school teachers and
that's completely gone.

Um, so first lesson,
don't build across a rupture.

The school has now
become a tourist attraction,

but Mike and Jing can see clues in
the landscape that suggest this
disaster could have been avoided.

The fact that beyond the school
buildings, the land is higher

and it may be there was an old scarp
here.

In the topography, you can see
the long-term effect of this fault

slicing straight up that valley
and giving that notch.

To be fair to the authorities,
there are many fault scarps
in these mountains

and they're very, very difficult
to find, we had only just begun
to find some of them,

it takes a long and sustained effort.

It took people in California
decades to map out the fault scarps
in any sort of precision.

There were fifteen million
people displaced by the earthquake.
The Chinese authorities

don't have time to wait until
they've mapped the precise location
of Sichuan's faults.

TRANSLATION: If someone shouts
"earthquake," put your hands
on your heads.

Hands on your heads
and hide under the desk.

The best that many schools
can do now

is simply rehearse for the moment
an earthquake strikes again.

Elsewhere in Sichuan,
they're rebuilding at a rapid rate.

But the vast majority of these new
homes won't be strong enough to
survive another major earthquake.

For Roger Bilham,
this is a problem that's endemic
throughout the developing world.

I can go here here here OK,
where my fingers stab the map,

there will be a magnitude seven
earthquake within,

you know, a few inches of it in the
next 30 years maybe in the
next year.

I've made a forecast that
it's possible right now

for one million people to be
killed by a single earthquake, OK?

Now, that's a terrible thing to say
and it's a thing that has

no precedent,
it's never happened in the past,

why can I make
such a crazy statement?

Because there are now cities of
eight and ten and twelve million
people along this earthquake belt

that have never been there
in the past and that knowledge
is sufficient, surely,

to drive those countries,
if they're responsible to mandate
earthquake resistance.

And it only costs
about another 10% more.

What it means is...

buying fewer guns and
better concrete instead.

Modern seismology has been
with us for over 100 years,

but scientists are still no closer
to predicting earthquakes.

However,
they haven't completely given up.

But where they once thought
it might be possible
to predict months in advance,

now, it's come down to
a matter of seconds.

We are prototyping
earthquake early warning, this is
also sometimes called now casting,

because it's not saying there's going
to be earthquake, its rather saying
an earthquake has already begun

and we're giving you that information
before the waves have travelled
from the fault to where you are.

The warning system will start
from stations like this,
located along the San Andreas Fault.

Instruments buried
deep underground will track

how much the fault is moving,
using high-precision GPS satellites.

The, er, antenna itself is inside
of this hemispherical shell and it's

constantly locked onto the radio
signals from the GPS satellites.

Each leg of this tripod
goes down about 30 feet,

ten metres into the ground and
firmly attached to the bedrock here.

But to have this system up and
running, these instruments need

to be able to feed back data the
moment the ground starts to shake.

In a big San Andreas earthquake,
this station would move
more than a metre

within a matter of less than ten
seconds, so other stations like this
positioned all along

the San Andreas Fault could actually
track the rupture as it's
progressing.

We're seeing it here coming up
the fault

and the red is where there's
a lot of damage, so we could have

used our stations in this area,
at this point, 20 seconds into
the earthquake.

Know that it's underway, we've got
a big earthquake started and send

that information to Los Angeles
that the earthquake's underway.

We could potentially get
up to a minute's warning.

You could hook this up to your
elevator and have your elevator moved
to the nearest floor and open

the doors, so people weren't
trapped for the next few days
after the power goes out.

We could ring an alarm in
operating room, so the surgeon pulls
the scalpel out of your chest.

You could ring an alarm where they're
handling toxic materials, so you're
not pouring out chlorine.

You could shut down critical computer
facilities, we could also be stopping
any rail lines, we could be flashing

the messages up on freeways, you
know, "earthquake coming, slow down."

Earthquake prediction doesn't come
any more high-tech than this,
and now casting

is not only possible,
it's surprisingly affordable.

Well, I think the public expects us
to be able to predict earthquakes,
and of course, we really can't.

But this is something that we can
do, we have the technology, we've
tested it, we've developed systems

that work and we know that we could
build an early-warning system,
at this point in time, we don't have

nearly the instrumentation
in place to be able to do that
kind of earthquake early warning,

some estimates, we think it could
cost about $100 million in all.

And the price tag we're looking at
for a big earthquake on the
San Andreas is, er, 200 billion

and up, so a $100 million system
to help reduce the damages
seems like a good investment.

This kind of early-warning system
might work for California one day,

but for most places in the world,
science's best answer to the threat
of earthquakes

is to construct
better buildings and map all the
faults in potential disaster zones.

It's really important to know where
these active faults are exactly,

so that at least if you can't
predict the earthquakes,

then we know where not to build.

But not everyone thinks that
prediction is totally dead,

there's still a sneaking hope
that someone, someday, may find
the Holy Grail of seismology.

Some seismologists would say, "No,
it's impossible and I'm not
willing to go that far,

"because we don't understand

"why exactly earthquakes happen,
so if we don't understand that, we
can't say they're not predictable."

It's an interesting challenge,
we might get closer to it, there
are obviously certain things

we're going to learn
and have learnt, maybe one day,

we'll get lucky and find that
we've been looking at
the wrong thing,

but right now, whatever
we do has resulted in, well,
I have to say failure, but you know,

we're trying, we're doing our bit,
we think we'll get there one day.