Episode 3 Storm Troupers: The Fight to Forecast the Weather

THUNDER RUMBLES It's variable, it's hard to predict,

it has a massive impact every hour of every day.

It is, of course, the weather.

I'm Alok Jha,

and I'm a science journalist.

I want to investigate how, through history,

people have tried to predict what the weather will do.

That's what this series is about.

The story of the extraordinary characters who took on

one of the hardest problems in science -

how to forecast the weather.

In this episode, the rise of the machines...

..how cutting-edge technology has helped some extraordinary men

and women transform meteorology.

The music lover who created the first computerised weather forecast.

The technocrat who took the UK Met Office to a brave new world.

And the mild-mannered mathematician who discovered chaos theory.

Since the Second World War, meteorology has become

one of the most highly-advanced scientific undertakings in history.

To me, this is one of the great achievements of modern science.

I compare it with mapping the human genome.

Today we see weather and climate as a giant global system.

Understanding what it has in store for us

is one of the biggest challenges humans face.

I'd like to know how much further we still have to go

before the Earth's weather and climate systems

reveal their deepest secrets.

Over the last 70 years,

weather forecasting has changed beyond recognition, becoming

one of the largest scientific and technological endeavours on Earth.

The story starts in the early 1950s, when the Met Office,

a small government department based in Dunstable, was the only

organisation charged with protecting the British public from weather.

Its staff used techniques refined during the Second World War,

gathering data from around 500 UK weather stations.

They drew charts of low pressure systems and weather fronts.

And then used their judgment to make a forecast.

In the 1950s, meteorologists were still using methods

that were really decades old.

So, science was being used,

they were capturing their ideas in terms of synoptic charts,

and these were very useful for forecasting the weather, say,

24 hours ahead, but it was still very much a hit and miss affair.

Things may have stayed like this for years. But, then...

LIGHTNING STRIKES

..in the most sudden and savage way possible,

the weather itself intervened.

I'm in the North Sea, off the coast of Essex.

It's calm here today, but on the night of January 31st, 1953,

a huge storm surge tore through this area, causing one of the worst

natural disasters in recent British history.

It was a perfect storm.

A low developed over the Atlantic,

which swept around the top of Scotland.

The air pressure dropped so much, the sea level rose by half a metre.

This coincided with an unusually high spring tide.

Then, vicious northerly winds tore down the east coast at 60 knots,

nearly 70mph.

That combination was deadly.

In some places, the waves were up to 20 feet high.

They battered this coastline, and overwhelmed sea defences.

The result was utter devastation.

Canvey Island in Essex was inundated.

Local councillor Ray Howard was 11 at the time.

We all went to bed early, being it was such a foul night.

But not realising that, soon after going to bed,

awoken by my sister, who come rushing into the room,

saying that there was water gushing down the street.

From Kent up to Lincolnshire, families like Ray's were stranded.

The water was getting deeper and deeper,

and then we saw boats coming down the street,

with Army personnel in, and, obviously,

they were shouting to say that they were going to start evacuating.

Across England, Scotland and the Netherlands,

32,000 people had to be evacuated.

And more than 2,000 died.

Communities would take years to recover.

We came back onto the island, and it was a different place.

I mean, the devastation those floods caused was unbelievable.

The thing that will always stick in my mind is

when I returned back to my school, and to see that boys and girls

that was in your class were not going to return.

So, how did the forecasters get it so wrong?

There would have been some indication. The observations

we collect would have allowed forecasters to see the fact

there was a depression, that it was moving, and, roughly,

what its track would be. And some sense that could result in floods.

However, there wasn't a warning system.

This is the big difference now.

There wasn't a way of communicating that risk to the public,

or to the police, or the fire folk

in a way that they could do something about it.

Action was urgently needed.

The government immediately began to rebuild Britain's flood defences.

And, for weather forecasters, the storm was a transformative moment.

The storm of 1953 shocked the Met Office into action.

The first thing they did was to institute

a nationwide warning service for storm tides.

But they knew they had to go further.

The first priority was to improve how to inform

the public about the weather.

Fortunately, the early 1950s saw the widespread adoption of television.

By 1953, more than three million Britons owned TVs.

To exploit this, the Met Office and the BBC set about making

forecasts more watchable by introducing on-screen presenters.

Well, as most of you will have realised,

it's been a very much better day today.

In the south of England, it has been for quite some time...

This is their first attempt at this new approach.

A trial, filmed in the autumn of 1953

with the Met Office's Jack Armstrong.

..however, it's not going to last very long.

As usual, there's another depression out in the Atlantic...

BBC television began regular broadcasts like this in early 1954.

By tomorrow, it is expected to be through into

the extreme north-east corner of Scotland by midday.

But the problem went beyond communication.

Forecasters in the 1950s could reliably predict

weather up to 24 hours ahead.

But the 1953 storm showed that this wasn't enough time

when warning large areas of the country.

Forecasters had to find a new way of doing things.

What they couldn't do was to really exploit the mathematical

and physical underpinnings of the subject.

So, this is it.

If you're going to break that barrier, go into two days,

three days, then you've got to be able to use the underpinning

mathematics in a more powerful way.

It had been known since the 19th century

that the weather's behaviour obeys just seven mathematical equations.

For a long time,

scientists had theorised they could be used to improve forecasts.

These seven equations tell us how wind speed and direction,

pressure, air temperature,

humidity, and density all interact with one another.

So, it's something you can write down on one side of A4,

and that is the basis of your model. It's as simple as that.

The difficulty is in then solving it.

In 1916, an English mathematician, Lewis Fry Richardson,

had tried to solve the equations of the weather.

He'd attempted this in his spare time between shifts

as an ambulance driver in World War I.

In doing so, he showed, in principle,

that mathematics could be used to predict the weather.

It was called numerical weather prediction, a technique that might

finally take the guesswork out of weather forecasting.

But the problem was the actual equations

are mind-bogglingly complicated.

Richardson laboured for two years, just to make a six-hour forecast.

So, the technique was theoretically sound but far from practical.

Then, some 30 years later, in the 1940s,

a new invention changed everything.

These are some of the first electronic computers.

They soon found many applications.

Everything from ballistics to accounting,

even airline reservations.

But the question was would they be powerful enough to produce

useful weather forecasts?

Very soon, meteorologists had hit a massive problem.

The weather is a complex and ever-changing interaction

between air temperature, pressure, humidity and cloud cover.

The computers of the 1940s just couldn't handle that much data.

Somebody had to make the equations of weather simple enough

to run on these early computers.

MUSIC: "Moonlight Sonata" by Ludwig van Beethoven

The person who did this was an extraordinary character.

An American mathematician called Jule Charney,

and his inspiration was music.

Charney once wrote to a friend to say that nature was a musician

more like Beethoven than Chopin.

What he meant was fewer higher-pitched, fiddly notes,

and more chords, bass lines and grand expressions.

Imagine the weather is like a musical instrument.

The high notes ...

HE PLAYS HIGH NOTES

..represent the ripples and disturbances that barely register.

Whereas the low notes...

HE PLAYS LOW NOTES

These are the things that really drive the weather system forward.

What Charney did was to come up with a mathematical framework

that allowed meteorologists to ignore and iron out

these little ripples and disturbances,

and focus instead on the things that really mattered.

It was a crucial step.

Charney had given computers a way to predict the movement of larger

weather systems without getting bogged down in details,

like small, local showers.

Or at least that was the theory.

Charney now had to make it work for real.

On Sunday, 5th March, 1950,

Charney kicked off the world's first numerical forecast.

He used a machine not too different from this one.

This is the Colossus at Bletchley Park, and it was designed

and built to crack German codes in the Second World War.

The machine Charney had was even bigger - the size of a room -

and it took an entire team of people,

working night and day, to keep it running.

To test Charney's ideas,

the team programmed their computer to recreate

the weather across North America on four separate days in early 1949.

They then compared the machine's predictions with the actual

weather on those days.

So, how did it do?

Well, the computer did correctly predict a low over America,

but it got some major details wrong.

The real low moved much faster

and covered a different area to the simulation.

As it happened, their first forecast,

and I quote, was "uniformly inaccurate".

But, by looking at that, they began to understand how to do computer

modelling, what the sensitivities were, how to develop them.

And it set the starting point, I think, for what has now

become the way that weather is forecast right around the world.

Charney's experiment ensured that modelling the weather

mathematically was the way forward.

Looking back, you have to admire these early

pioneers for their scientific ambition.

I find it remarkable that those people who began to programme

these equations into computers and have the bravery, almost,

to try and model something as, as messy as the weather

in computers, using equations.

The fact that they tried, I think is incredible.

The pioneers of computerised weather forecasting

would need all the bravery and determination they could muster.

In Britain, the story of how computers became central

to our meteorological science was as much down to

individual personalities as it was to raw technical innovation.

To tell it, I've come to one of the world's most advanced

weather measurement facilities, and it's in Hampshire.

This is the Chilbolton Observatory.

It's the biggest steerable radar antenna in the world

being used for weather forecasting.

It's gathering information about clouds.

It's helping to steer meteorological flights,

and it's even tracking weather satellites.

If there's a symbol for how technological

weather forecasting's become, this is it.

It seems obvious now that something as complex as the weather

would need equipment on this scale in order to keep track of it.

But it wasn't always the case.

When this kind of technology was being developed

in the 1950s and 60s,

many meteorologists were unconvinced

that it would ever be useful to them.

In those decades, a real tension developed between man and machine.

Many were unconvinced that the raw calculating skills of computers

would ever replace good old-fashioned human judgment.

There was actually a great deal of scepticism in the community

that even with modern electronic computers,

they could actually do the calculations in a reliable way.

In other words, they could see that there were plenty of pitfalls,

plenty of opportunities for calculations to go wrong.

And, therefore, they didn't really have the confidence

that numerical techniques would actually be superior

to a human forecaster.

Overcoming this attitude would need a strong character.

Someone who was both ambitious and a top-flight scientist.

This is John Mason, a physicist with a long fascination for weather.

He was appointed director-general of the Met Office in 1965,

aged just 42.

Dr Stan Cornford was there when Mason first took over.

He didn't seem an attractive person at that stage.

But, in fact, he was a very able man

and he didn't underestimate his abilities.

Mason knew the world was changing,

the economy booming.

Growing industries like aviation

demanded ever more accurate and detailed forecasts.

And he was convinced that computers were the best way to provide them.

He ensured that we could use the best available scientific tools, I suppose.

And, in particular, always making sure that we had the biggest

and best computer that was possible.

Mason was a very firm part of that

and his leadership helped make it possible.

Mason was a man in a hurry.

With older and wiser colleagues telling him

that computers weren't really up to the task of weather forecasting,

he decided to force the issue.

In November 1965,

he organised the Met Office's first ever press conference.

There, he made an announcement

from which there could be no turning back.

At the press conference,

John Mason took the biggest gamble of his career.

He said that from then on, all Met Office predictions

would be based on computer calculations,

rather than just human skill and experience.

Brimming with confidence that many of his colleagues didn't share,

Mason promised the assembled press something incredible.

Two computer-based weather forecasts every day,

that he said would be more accurate than anything they'd seen before.

Further raising the stakes,

Mason handed out copies to each of the journalists present.

And here it is.

The very first operational weather chart made by a computer.

It might not look like much,

but this is a landmark in modern weather forecasting.

The computer-generated chart predicted calm and settled weather

for the next day, for most of the country.

It was vital for both Mason's reputation,

and the Met Office's future as a hi-tech organisation,

that the actual weather matched this forecast.

Fortunately, the Met office computer got it right.

The forecast that it made for November 3, 1965

was an excellent match for what actually happened in the real world.

Journalists, even politicians were impressed.

That success allowed Mason to secure the funding he needed

to remake the Met Office in his hi-tech vision.

Mason's timing was impeccable.

He put his faith in computers at a time

when another spectacular breakthrough was about to take

their ability to forecast the weather to new heights.

Tiros, meaning Television And Infrared Observation Satellite,

is more than just a dream.

This experimental weather satellite means that man's vision

is no longer limited to looking up at the clouds gathering above him.

That was really the game changing moment.

That was when scientists realised the potential,

that if we could make measurements from space in a controlled way,

ie with a satellite in a controlled orbit of the Earth,

then that really would give us this all-encompassing global picture

of what the atmosphere's doing now.

These are some of the first images of our planet from space,

showing the global weather system.

This one, taken of Hurricane Esther in 1961,

was the first direct evidence for how such storms develop.

Satellites greatly improved computer-based forecasting,

which needs the best information about how the weather is behaving now

in order to calculate what it'll do next.

Your computer models,

you have to feed them with what we call the initial conditions.

So, the conditions of the atmosphere now,

you need to know that perfectly over the whole globe.

Frankly, when the satellites came,

that was really a revolution

because that's what gave you these global pictures

that we were waiting for.

From the mid-1960s,

satellites and computers have given us a powerful way

to calculate how weather systems evolve

from their initial atmospheric conditions.

It's perhaps the biggest reason

why we can have such a close relationship with the weather today.

From what we should wear,

to whether we should put the heating on.

From what we choose to eat and drink,

to when we leave the house.

Even how we choose to get around.

Nowadays, it's amazing how accessible and how accurate

the short-term weather forecast has become.

RADIO: Northwest Fitzroy...

But despite this, the last 50 years have not been smooth sailing.

..occasional rain, moderate or good.

Occasionally poor.

..Irish Sea.

Southwest, five to...

We all take weather forecasts for granted.

But they're by no means perfect.

In fact, there are very recent examples of when they've gone drastically wrong.

Good afternoon to you. Earlier on today,

apparently a woman rang the BBC and said she heard there was a hurricane on the way.

If you're watching, don't worry, there isn't. But having said that...

This is probably the most infamous weather forecast

ever broadcast in the UK.

On 15 October 1987,

the BBC's Michael Fish joked with viewers about a hurricane

that was rumoured to be heading towards Britain.

He was dismissive of the idea.

Most of the strong winds, incidentally,

will be down over Spain and across into France as well.

But there's a vicious-looking area

of low pressure on our doorstep nevertheless,

around about the Brittany area.

And that is going to head across the southeastern corner of the country

bringing, if nothing else, a lot of rain with it.

But what happened next?

Well, we all know what happened next.

Good afternoon.

The worst storms for hundreds of years hit the south of England

earlier this morning, killing a dozen people

and bringing the whole south-east to a halt.

Gales of over 100mph smashed buildings

and caused millions of pounds worth of damage.

There was no warning.

The weathermen were caught with their forecasts down.

In the early hours of the 16th of October,

the great storm of 1987 smashed into England as she slept.

Winds of 122mph ripped across the south-east of the country.

Roofs were torn from homes, power lines were down,

road and rail networks were blocked,

and in all, 18 people lost their lives.

Most of the damage was done by falling trees,

even huge mature ones like this

were uprooted by the hurricane force winds.

Here on Hampstead Heath, many of them still lie

exactly as they fell on that stormy morning.

Not surprisingly, people were furious.

Well, joining me now from the London Weather Centre,

is the BBC's weather forecaster Ian McCaskill.

Well, Ian, you chaps were a fat lot of good last night.

Well, we have been forecasting

high winds and gales relentlessly since Sunday. We admit...

I admit we weren't forecasting hurricane force winds

and that's what we got and that's what we will get

maybe once every 50 years, maybe once in a lifetime.

By this time, meteorologists understood

how a storm of this size develops.

Fish and his colleagues had even forecast strong winds, as they

recognised a massive low-pressure system moving in from the Atlantic.

But the crucial moment when it suddenly intensified just

hours before making land, happened in a narrow

band of the Atlantic between any observation ships or weather buoys.

It's impossible to know whether any lives could have been saved

or damage avoided if the forecast had been any different.

But what we do know is the whole episode shook

confidence in the Met Office.

This is the organisation that was meant to protect us,

to warn us against the worst effects of the weather.

Just as in 1953,

this very public disaster was a wake-up call for forecasters.

Was that a bad moment for the Met Office?

Well, actually, on the day, probably yes.

But in the longer term, no, it wasn't.

It was a really good thing to have happened for us, because it

made us think very seriously about how we can do this better.

What science do we need to invest in?

How do we transform our forecasting system

so that this doesn't happen again?

As a first step, an internal Met Office

inquiry called for an increase in the quality

and quantity of weather observations off the south-west coast of England.

But that's not all.

The Michael Fish broadcast was based on a computer model

and it forced the Met Office to look again at how

they used computers to make forecasts.

Computer models, based on the strict rules of maths and physics,

were supposed to take the guesswork out of forecasting.

How could they possibly have failed so dramatically?

The answer lay in the fact that the fundamental mathematical

equations of the weather contained a truth about nature that

no-one wanted to confront.

Our weather forecasting model is based on some mathematics

that has a rather special property to it,

and it's a property that makes weather very interesting,

and now and again, makes weather forecasting very difficult.

What they had to face up to, after the '87 storm, was that there was

something else there which just couldn't be left

out of the forecasting picture any longer.

In fact, the way scientists understood the world had to

be completely revolutionised.

By the start of the 20th century,

scientists saw the world like a clockwork toy.

If you understood the parts of the system,

and how those parts interacted with each other,

you could predict everything about that system's future,

whether it's the orbit of the planets around the Sun or

a ball flying through the air.

And weather prediction was going along similar lines.

If you could understand how the air moved,

then you could predict the weather.

Unfortunately, a discovery in the mid-20th century shattered

those dreams of absolute predictability.

And it's something you've probably heard of.

Chaos theory.

This told scientists that however perfectly they understood a system,

making accurate predictions about its future were almost impossible.

This completely changed how they saw the world, and for weather

forecasters, it meant that uncertainty and unpredictability

had to be brought right to the heart of what they were doing.

It all started with these unremarkable looking squiggles.

But make no mistake,

this is one of the most significant images in the history of science.

The first-ever proof of chaos theory.

To tell the story, we have to go back in time to 1965, to the USA

and the Massachusetts Institute of Technology.

Here, a mild-mannered meteorology professor called

Edward Lorenz worked with early computers, generating weather

models as a way to combine his two great loves - maths and weather.

Edward Lorenz had a set-up a bit like this one.

He'd input data here about the weather - things like temperature

or air pressure - the computer would take those numbers, run them

forward in time and then print out its results here as a graph.

That, in essence, was the weather forecast.

Lorenz's model was incredibly basic.

Using just three mathematical equations to simulate

how his initial conditions might change in a real weather system.

But from his early results, it looked as though it was doing a good

job, outputting weather patterns that seemed to mimic the real thing.

So far, so good, but Lorenz wanted to check the simulation,

so he decided to run the whole thing again, using the same starting data.

All he wanted to do was repeat the results from his first run

to prove that his model was accurate and reliable.

But what he saw shocked him.

Lorenz expected that both his runs would produce identical graphs,

and at the start, that's exactly what they did.

But very soon, the curves began to diverge,

and by the time you get over here,

they are two separate curves altogether, and that made no sense.

Why would the computer plot two different curves from the same

starting data?

It was a mystery.

The model itself hadn't changed, and as far as he knew, Lorenz

had put in his data exactly as he had recorded it from the first run.

Why on earth was the second curve so different to the first?

The answer turned out to be remarkably simple.

Lorenz's computer stored its data to six decimal places,

but to save time, in a second run,

Lorenz had only input the data to three decimal places.

It was a small change, a fraction of a percent,

but it made all the difference

and it accounted for those wildly different graphs that he saw.

This is the key idea in chaos theory -

an immeasurably small difference at the start of a process can

have huge consequences to how it ends up,

and this effectively makes that process unpredictable.

One of the hardest ideas to get your head around in science

is that while some systems are inherently predictable...

..others seem inherently unpredictable.

These unlikely visual aids represent the key

to explaining what Lorenz saw in his results.

And it's all to do with their shape.

This football is a sphere, and when I drop it on the ground,

I know where it's going to end up.

Even if I change the starting conditions and drop

it from higher or at a different angle, I can predict its path.

Things are very different for this rugby ball.

When I drop this, I honestly don't know where it will end up.

Now, every time I drop this ball,

it went in a completely different direction, which makes no sense,

because both of these balls operate under the same laws of physics.

The only difference really is the shape,

but that difference turns out to be crucial.

Every time I drop this ball, even the very slight

differences in the starting conditions, accidental ones

really, will make huge differences in how the ball bounces around.

And this is how the weather works.

When Lorenz was making his computer models, he thought

that weather operated a bit like this football - predictably.

In fact, weather is much more unpredictable,

like this rugby ball.

At its core, this is the very essence of chaos,

where a tiny error at the outset doesn't make for a tiny difference

in the outcome, it makes for a different outcome entirely.

In the case of the 1987 storm,

the tiny error came from missed observations of the atmospheric

conditions in a small part of the North Atlantic.

So what this reveals is that actually just missing one

bit of crucial information can, now and again,

and it should be stressed, now and again, make a big difference.

And that is the hallmark of chaotic behaviour in the atmosphere.

The lesson from the 1987 storm was clear -

computerised weather prediction had to be reformed,

so it could deal with the chaos inherent in weather.

This is the European Centre for Medium-Range Weather Forecasts

near Reading.

In the 1980s, Tim Palmer worked with a team here to come up with

a new style of forecasting to combat chaos.

-Hello, Tim.

-Hi, Alok, how are you doing?

-This looks incredible.

Welcome to the European Weather Centre.

So just tell me what we're looking at here.

We're looking at an ensemble forecast,

this is a modern way of doing weather forecasting.

An ensemble isn't just one forecast, but 50.

Forecasters give their computer a range of starting conditions,

rather than a single best guess, like they did before.

What we're seeing here are the 50 individual weather forecasts

that make up an ensemble.

They are run from almost but not quite identical starting

conditions and then we move these 50 models forward in time to next week,

and then we look to see, do all these 50 forecasts stick together?

In which case, one can be quite confident about what the future

weather is going to be, or do they all diverge and do their own thing?

This is kind of shown more clearly here, where we've zoomed in on just

a few of these individual members, if you like, forecasts.

So for example, this one would be a rather unpleasant day

across much of the UK - blustery, cold, very windy type of day,

not the day to have a picnic, by any means.

But just move one member over, so 23 to 24,

and this actually is meteorologically quite different.

There's very little pressure gradient across the UK.

It's sort of a reasonably nice day.

There might be a shower or something, but it could well be a very pleasant day.

So in each of these predictions, you're getting a spread

of how the weather might change over the next 12 hours or seven days?

That's right. We're getting basically 50 plausible,

equally plausible, equally likely, a priori estimates of the weather.

The clever thing about this ensemble method is that

if the 50 forecasts are all similar,

then we can be confident about what the weather will do.

But if they are very different,

we have to accept that we can't be sure what will happen.

So this tool allows us to estimate that sort of degree of confidence

or degree of predictability ahead of time,

so the forecasters these days on the TV

will often talk, actually, about uncertainty.

They know it's uncertain, because they've looked at these 50 forecasts

and they can see, in half of them

the low-pressure tracks up towards Scotland,

but in the other half, it tracks across southern England.

And built into this system is a way of preventing mistakes,

as with the storm of 1987.

Suppose there are just three or four of the forecasts which have an

incredibly violent and intense storm,

you may not want to ignore that,

and if those are situations where people's lives are risks,

for example, then that is a dangerous situation to be in.

So pretty much all around the world now,

this is the technique that is used,

because it provides not only the most likely situation,

but it provides probabilities of extreme weather

and it provides estimates of confidence.

Ensemble forecasting really came into its own

when the St Jude's Day storm struck on October 28th 2013.

It produced violent winds that were

only slightly less powerful than the 1987 storm,

and it took a similar path across the south of England.

But the big difference was that in 2013, we were ready for it.

Warnings about today's storm were first given a week ago,

giving people and the emergency services time to prepare,

unlike the great storm of 1987,

which took weather forecasters by surprise.

A week's warning for the St Jude's Day storm

showed that satellites, computer models

-and ensemble forecasting

-have

-made a difference.

Let's go across and take a look at today's chart.

Compared to how they were in the 1950s,

our forecasts have improved enormously.

Our five-day forecasts today

are as good as the one-day forecasts back then.

Well, that's the forecast. Bye-bye for now.

But what if we wanted to look further ahead?

How good are we at predicting what the coming season will be like?

Hello again. Let's begin with what

should be good news for most of us.

The Met Office have issued a long-range forecast for the summer.

Here's the headline -

we're more likely to need a barbecue than a brolly.

In April 2009, the Met Office prediction

of a barbecue summer went badly wrong...

There will be some rain at times,

but we're not expecting a wash-out, by any means.

..As what actually happened was one of the wettest summers on record.

And then, in the same year,

the Met Office said the chances of a cold winter in the UK were only 20%.

In reality, the entire island was frozen solid.

Since these events, the Met Office has stopped issuing seasonal

forecasts, replacing them with monthly outlooks.

The demand and interest for forecast weeks and months

and years ahead is huge.

The science of that, arguably, is very much in its infancy.

We are beginning to understand the processes that affect

weather in those weeks and months ahead.

Long-term forecasts are hard,

because the effects of chaos increase with time.

A ten-day forecast isn't simply twice as hard as a five day one,

it can be orders of magnitude more difficult.

To see why, I want to show you a simple experiment with a twist.

This is a simple pendulum.

You'll have seen these in grandfather clocks or even

demos in science lessons at school.

And they're very straightforward.

Even I could write you the mathematical equations

that govern how this works, and even tell you when it stops.

But it doesn't take much to turn this very simple system

into something completely unpredictable.

This is a double pendulum.

It doesn't look hugely different to what we had before,

but watch what happens when I swing it.

At first, it's all very normal, but very soon,

it starts to behave completely erratically.

Now, physicists are terrified of the double pendulum,

because you can write the maths for this, but you can't solve anything.

You can't predict what's going to happen.

And you can see here, this completely erratic motion.

And it's because there's two parts to this system now.

The top pendulum affects the motion of the bottom one,

which in turn, affects the motion of the top one.

And this feedback causes all sorts of headaches

if you want to know what's going to happen next.

Weather is full of these feedback loops.

Factors like humidity, temperature

and air pressure all interact with each other, as the trillions

of atoms that make up our atmosphere collide on a global scale.

So a small change in one variable can be multiplied across the others.

This means the longer a system runs,

the more unpredictable it can become.

This chaotic behaviour is the reason why however good our

observational technology is, however sophisticated our computer models,

precise weather forecasts beyond a few weeks

are still out of our grasp.

Chaos has always been there, but we no longer ignore it.

-Meteorologists don't speak about what

-will

-happen,

-but what

-might

-happen.

And for businesses, who have to calculate

risks in the long-term, knowing

the probability of what the weather might do is crucial information.

From the energy sector to supermarkets to civil aviation,

forecasting is vital to the economy.

And weather prediction itself is big business today,

with more private companies

than ever before providing thousands of forecasts every day.

All this is possible

because of our understanding of the chaotic nature of our atmosphere.

But there's another frontier to atmospheric science

beyond just predicting the weather.

And it's become one of the most important subjects

in modern science.

Predicting the climate.

In simple terms, climate is just the average of how the weather

behaves over the very long time, so whereas in a weather prediction,

we might be looking a few days or weeks ahead,

in a climate prediction,

you're looking decades, even centuries ahead.

It's not easy to do and there are a lot of uncertainties involved,

but it's incredibly important,

and we've been trying to do it for much longer than you might think.

Climate science hinges on understanding not just

how our atmosphere moves around the planet, but what it's made of

and how it interacts with the oceans and the land below.

Any changes on this scale need decades to take effect.

But we have been able to make predictions about their impact.

The first breakthrough was the realisation that carbon dioxide,

released by burning fossil fuels, might cause the planet to warm up.

In the '70s, I wrote one of the first papers

on what the effects of doubling

the carbon dioxide concentration in the atmosphere might be.

At that time, I thought of this as a bit of an academic exercise,

but of course, now, here we are,

it being one of the defining problems of the 21st century.

This effect has been on the world's agenda for some time now.

But Julia was not the first person to discover it.

She was picking up on work done much earlier.

Our knowledge of man-made climate change dates back to

an experiment done in the mid-19th century.

I've asked chemist Professor Andrea Sella to recreate it for me.

-Andrea, hello. How are you?

-Hey, Alok, good to see you.

What is this remarkable set-up you've got here? It's incredible.

Well, this is a recreation of one of the great demonstrations that a man

called John Tyndall did at the Royal Institution in the mid-19th century.

So I can see lenses, there is a light source.

What does this show me about anything?

Well, what Tyndall was trying to understand really was what

was it in the atmosphere that absorbed different types of light?

So what we've got is a very, very intense light source,

and it's giving out a combination of two things.

One is that it's giving out loads of visible light, but the other thing

is it's also producing what was then called radiant heat.

In other words, you could feel the warmth coming from that lamp.

This light represents two things - the Sun and the warmth that's

radiated back into the atmosphere from the Earth's surface.

So what Tyndall showed was the fact that it was actually heat

radiation, if you will, which could actually be focused down.

And I'm going to show you that, with a little piece of guncotton.

What I'm going to do is I'm going

to put it actually into the beam here and we're going to see.

What am I looking for?

Well, something might actually happen if the whole thing gets hot.

Let's just see if I can get it in the right spot.

-It's incredibly bright right now.

-It's quite tricky.

-Ooh!

-Ooh!

-OK.

-OK. That was obvious.

So clearly, this stuff has gotten hot enough to actually ignite.

Now... Yeah, it does smell a little bit.

So what Tyndall did then was to actually systematically go

through the various components of the atmosphere and of course

those included nitrogen and oxygen and water vapour and so on.

And then he actually put carbon dioxide in between.

So in here, you've got carbon dioxide.

I've taken a little bit of carbon dioxide in the form of dry ice.

I've left it inside so that it evaporates.

So that tank is completely full.

And what I'm going to do is I'm actually going to put a little

bit of the same cotton here and I'm going to place it at the focus.

-Are we expecting something similar?

-Well, let's...let's see.

Let's see what happens.

So, I've got it at the focus,

-you can see the way it has brightened up.

-Still bright.

It's still bright.

OK, this is less exciting, Andrea.

Well...exactly.

The incredible thing is that you could leave this here

for a very long time, and it just doesn't catch fire.

The reason is, because in fact, this tank of CO2

is acting as a kind of filter along the way.

So this is absorbing the heat that would normally have passed

-straight through.

-It's absorbing that energy.

And so the result is that that is going to get just a little

bit warmer, but it prevents anything happening at the far end.

'What's happening on this desktop demonstrates the warming

'effect of carbon dioxide in our atmosphere.'

But if you have more carbon dioxide in the atmosphere,

the inevitable consequence is that the Earth's temperature will get

higher and higher slowly.

Exactly, and way back in 1855, John Tyndall suggested that

if we continued to burn coal,

which was of course the fuel of the Industrial Revolution,

that the world would become warmer as a result.

He suggested, essentially, what we know of now as global warming.

The extraordinary thing is that we've known about this stuff

for something like 170 years.

The implications of this experiment are still being wrestled with.

In fact, understanding how our climate is changing

and what the impacts might be is the biggest challenge

faced by meteorologists today,

shaping a new approach to the whole field.

So, whereas I started my career having to understand

meteorology and atmospheric motions and atmospheric physics,

I now need to understand about ocean physics,

ocean biogeochemistry, how plants grow, how sea ice,

particularly in the Arctic, how that forms and melts.

All those things we need to understand now,

if we're going to say useful things, robust things,

about how our climate will evolve as we go into a warming world.

In the UK, many institutions share this huge task.

Collaborating on projects

like the Facility for Airborne Atmospheric Measurements,

which operates dozens of flights a year to study our climate.

Today, scientists on board will test an instrument that will

measure aspects of our planet in an entirely new way.

It's really exciting to be on these flights.

It's a new instrument and, you know, we've been working on it

for a very long time, so the chance now to see

it flying and measuring real data is really exciting to me, personally.

This brand-new technology is the latest tool to carry on Tyndall's

early experiments into how our atmosphere absorbs light and heat.

It's designed to measure a key factor in how much radiation

gets trapped in the different layers of our atmosphere.

Each of these channels is looking at a part

of the spectrum where the atmosphere is different transparency,

so some of them see quite well right through the atmosphere

and some of them only see the bit fairly close

to where you are at the moment, and that's why

we've got this spread in the amount that's coming down.

So this one here isn't seeing very far away from us,

so it's getting quite high amounts of thermal radiation.

What we're looking at is quite warm,

whereas this one here is basically seeing almost all the way

through to space and space is very cold,

so this is giving us temperatures of about -250 Celsius,

so not much radiation coming down there at all.

Of particular interest to Stuart is the effect of clouds.

Some contain a high amount of ice,

and these reflect the Sun's rays back into space.

This instrument will measure a new thing,

it's going to measure cloud ice.

So once it's up on the satellite, it will be giving us

daily global measurements of cloud ice content, which is

a really important thing to get right in our climate models,

because it's got a big impact on how much of the solar radiation

gets down to the surface and how much of it

gets reflected back up into space.

Once tested on the plane,

Stuart's sensor will be launched into space on a satellite,

adding to the vast network of weather observation systems

that circle our planet.

When the very first satellite was launched,

the models were pretty highly developed, we knew quite

a lot about the atmosphere, but we didn't have many observations.

We're now in the situation where one could argue

we have more observations than we really know what to do with.

In the years immediately after the Second World War, Met Office

forecasters received a few hundred observations per day.

Now, thanks to everything from aircraft

to a global network of satellites, they receive a staggering

106 million weather observations every single day.

Bringing those observations together, bringing those together

in a coherent way, that's the challenge for science at the moment.

Making sense of all this data, to figure out what it can tell us

about our future, is an enormous task.

And it's one that the Met Office have to take on

in order to uphold their commitment

to protecting the British public from danger.

That means building this,

one of the two biggest weather computers in the world.

It's the only one that can simulate both short-term weather

and long-term climate simultaneously.

On every scale, this computer is enormous.

It takes up two rooms, each one the size of a football pitch.

It weighs the same as 11 double-decker buses,

and it can compute 23,000 trillion calculations per second,

but that's what you need if you want to simulate the entire Earth

and all its weather systems.

The computer code running in this machine

has taken 100 scientists ten years to write.

It's a virtual planet Earth, complete with simulated weather

that churns about the atmosphere with astonishing accuracy.

This single tool is both a magnifying glass and crystal ball.

Up to a week in advance, it can show us what the weather

will do in Britain down to resolution of one square kilometre.

And looking decades into the future, it can give predictions for

how our climate will behave across the globe.

When we think about what our models enable us to do,

that we can actually simulate what we see going on outside,

see our weather being simulated, to me,

this is one of the greatest achievements of modern science.

It's probably one of the unsung

great achievements of modern science.

Just because it's a computer code and therefore not visible,

doesn't mean it isn't something that is a huge success.

I compare it a bit with the human genome, mapping the human genome.

It's a similar achievement.

This is a remarkable coming together of science

and encapsulating all our understanding of that

science in these codes and producing something, a simulation that

looks so remarkably like reality, it's an extraordinary achievement.

This giant machine encapsulates everything meteorologists

have learnt throughout the history of their science.

Hard-earned insights gleaned from tragedies and triumphs alike.

The early pioneers wanted to see what the weather

had in store for us just over the horizon.

And now, thanks to their legacy,

we have the tools to see a whole planet at a glance...

..and look hundreds of years into the future.

We've been studying the weather for more than 150 years,

and in that time, we have learned how to understand

and predict it, but if the weather has taught us anything at all,

it's that this is a mysterious and intriguing force of nature,

always slightly out of our grasp.

We need to ask deeper questions, probe even further,

if we're going to keep up to pace with what it has in store for us.