Nine Days

---

By Justin RowlattClimate editor, BBC News

Data -tip='diger' Data -kimlik='4131976'>Artificial Intelligence could save lives by warning where a hurricane will hit land Much sooner than traditional forecasting systems, researchers say.

A new AI tool from Google Deep Mind predicted where September's hurricane Lee would make landfall in Canada Data -tip='diger' Data -kimlik='4131976'>Three Days ahead of existing methods.

Weather forecasts have become Much more accurate over the decades.

But AI's speed and ability to analyse past events to make predictions make it a game-changer, say scientists.

An accurate Data -tip='diger' Data -kimlik='4131976'>Weather Forecast is useful to tell you what to wear when you go out in Data -tip='diger' Data -kimlik='4131976'>The Morning But - Much more importantly - Can forewarn us of Data -tip='diger' Data -kimlik='4131976'>Extreme Weather like storms, floods and heatwaves, giving communities Data -tip='diger' Data -kimlik='4131976'>Crucial Time to prepare.

However, traditional weather forecasts take vast amounts of computing power.

They involve creating estimates of hundreds of factors including air pressure, temperature, wind speeds and humidity at different levels of the atmosphere around the globe.

A new AI tool called GraphCast created by Google Deep Mind outperforms the European Medium Range Weather Forecasting Data -tip='diger' Data -kimlik='4131976'>Model - One Data -tip='diger' Data -kimlik='4131976'>The Best in Data -tip='diger' Data -kimlik='4131976'>The World - on Data -tip='diger' Data -kimlik='4131976'>More Than 90% of those factors, according to a peer-reviewed paper published by Deep Mind in the journal Science.

GraphCast produces its forecasts in less than a minute, using a fraction of the computing power of traditional forecasting methods because it takes a very different approach.

Traditional weather forecasting involves taking measurements of what is happening in the atmosphere right now.

Data -tip='diger' Data -kimlik='4131976'>The Best models take in hundreds of millions of readings from around Data -tip='diger' Data -kimlik='4131976'>The World Data -tip='diger' Data -kimlik='4131976'>Every Day .

These come from a huge range of sources including weather stations, satellites, balloons sent up in the atmosphere, buoys in Data -tip='diger' Data -kimlik='4131976'>The Ocean - Even readings taken by sensors on the noses of commercial jet planes.

" We then use our Data -tip='diger' Data -kimlik='4131976'>Model to select which are going to be Data -tip='diger' Data -kimlik='4131976'>The Most important, " explains Matthew Chantry, of the European Centre for Medium Range Weather Forecasting (ECMRWF) who says about 10 million of the measurements will be used for One of its forecasts.

This ocean of Data is fed into a supercomputer to be processed by programmes which Can do quadrillions (a thousand trillion) of calculations every second. These use complex equations to simulate what happens in Data -tip='diger' Data -kimlik='4131976'>The Earth 's atmosphere to predict how Data -tip='diger' Data -kimlik='4131976'>The Weather will change and evolve over time.

This method has been extraordinarily successful. As Data -tip='diger' Data -kimlik='4131976'>The Models have improved and Data -tip='diger' Data -kimlik='4131976'>The Computers have got more powerful over the decades, weather forecasts have got significantly more accurate.

But these numerical weather prediction (NWP) models, as they are known, take vast amounts of computer resources, using some of the biggest supercomputers in Data -tip='diger' Data -kimlik='4131976'>The World and typically take hours to produce their forecasts.

A new approach

AI shortcuts Much of this effort. It does not try to Data -tip='diger' Data -kimlik='4131976'>Model how Data -tip='diger' Data -kimlik='4131976'>The World works.

Instead, GraphCast uses Data -tip='diger' Data -kimlik='4131976'>Machine Learning to digest vast quantities of historical Data - including the output of the ECMRWF Data -tip='diger' Data -kimlik='4131976'>Model - to learn how weather patterns evolve.

It uses this knowledge to predict how Data -tip='diger' Data -kimlik='4131976'>The Weather now is likely to change in Data -tip='diger' Data -kimlik='4131976'>The Future .

And it is proving very effective.

" The main advantage of this AI approach is that it's extremely accurate, " said Remy Lam of Google Deep Mind, who helped create Data -tip='diger' Data -kimlik='4131976'>The Weather tool.

" It learns from decades of Data and is able to be more accurate than the industry gold standard, " he says.

And, because it does not try to solve complex equations, it Can make its forecasts very quickly and using Much less computing power.

GraphCast's forecasts are not as detailed as those produced by the ECMRWF But it is better at predicting severe events like extreme temperatures and at tracking Data -tip='diger' Data -kimlik='4131976'>The Path of big storms.

It accurately predicted where Hurricane Lee, a storm that hit Data -tip='diger' Data -kimlik='4131976'>The Atlantic coast of the US and Canada in September, would make landfall, for example.

Deep Mind's AI tool predicted its path Data -tip='diger' Data -kimlik='4131976'>Nine Days ahead, the ECMRWF only managed Data -tip='diger' Data -kimlik='4131976'>Six Days ahead.

But the success of GraphCast does not mean we Can shut down the supercomputers and rely on AI instead.

Even Remy Lam from Google Deep Mind says that will not happen.

'Huge leaps forward'

" Data -tip='diger' Data -kimlik='4131976'>We Are standing on the shoulders Data -tip='diger' Data -kimlik='4131976'>Of Giants to build those models" he says.

Rather than replacing traditional weather forecasts AI models will complement them, he believes.

" AI models are trained from Data and that Data is generated by traditional approaches, so we still need the traditional approach to gather Data to train the Data -tip='diger' Data -kimlik='4131976'>Model , " says Mr Lam.

GraphCast is open source which means Google Deep Mind is sharing Data -tip='diger' Data -kimlik='4131976'>The Details of the design so anyone Can use the technology.

Many technology companies and weather and climate organisations around Data -tip='diger' Data -kimlik='4131976'>The World are designing their own AI weather prediction tools.

The Data -tip='diger' Data -kimlik='4131976'>Met Office , the UK's national weather service, is working with the Turing Institute, the country's Data Data -tip='diger' Data -kimlik='4131976'>Science Centre to explore the potential for AI to improve weather forecasting, for example.

" Weather forecasts derived from Data -tip='diger' Data -kimlik='4131976'>Artificial Intelligence and Data -tip='diger' Data -kimlik='4131976'>Machine Learning are taking huge leaps forward, " acknowledges Prof Simon Vosper, the Data -tip='diger' Data -kimlik='4131976'>Met Office 's Director of Science.

But he warns Data -tip='diger' Data -kimlik='4131976'>Climate Change will limit the predictive power of AI based tools.

" Data -tip='diger' Data -kimlik='4131976'>We Are seeing new climate-related weather extremes, such as Data -tip='diger' Data -kimlik='4131976'>Last Year 's 40C temperatures in the UK that would haven't been realised in former times" says Prof Vosper.

Data -tip='diger' Data -kimlik='4131976'>The Way Data -tip='diger' Data -kimlik='4131976'>Extreme Weather systems evolve may also be changing.

Hurricane Otis rapidly intensified from a tropical storm into Data -tip='diger' Data -kimlik='4131976'>The Strongest Data -tip='diger' Data -kimlik='4131976'>Category 5 hurricane over just 24 hours in October before making devastating landfall on the coast of southern Mexico.

Climate scientists warn rising ocean temperatures are likely to make this process of rapid intensification of storms more common.

" So it is fair to question whether AI-based systems are able to Data -tip='diger' Data -kimlik='4131976'>Pick Up new extremes if these systems have only been 'trained' on previous weather conditions, " suggests Prof Vosper.

Related Topics

Source of news: bbc.com