The extreme science of climate forecasting | FT Climate Capital

You can enable subtitles (captions) in the video player

5, 4, 3, 2, 1, 0. And lift-off of Sentinel-6.

The quality and quantity of data beaming into the world's weather centres has never been better.

The technology we have in space now is just incredible.

In essence, we're facing a tsunami of data.

It's never been more important to measure and predict the climate.

Constellations of satellites and batteries of instruments are collecting huge volumes of data on land and at sea. Hundreds of millions of observations every day.

Leading climate agencies use supercomputers and complex mathematical models to process this data, and the results are unanimous: record high surface and sea temperatures well above the pre-industrial average.

So when we look at the overall trends of different data-producing centres, they're all showing the same trend through time. So we all agree that 2023 was the warmest by far.

Given that we say that every fraction of a degree matters, how can we be sure that the data we're collecting is accurate given the variables?

Because we've all got slightly different methodologies. The important thing to remember is even if there's tiny divergence in the final number, the trend is the same through time.

Everyone is using different statistical, mathematical techniques to put all that data together, and that independent working is a good check because the answers that come out are very close.

The sharing of one particular data-gathering technology has been a game changer for climate scientists.

We have about 100 dedicated meteorological satellite instruments on board a variety of spacecraft right now above our heads.

Operating across a range of wavelengths, satellites are constantly measuring what would otherwise be hidden. From methane emissions to the thickness of sea ice.

But we have observation systems operating in the infrared, in the microwave. We also have things called active sensors, which are firing lasers down into the atmosphere and looking how the laser signal is scattered. And we have radar systems, which are beaming down from satellites onto the ocean surface and the land surface, and we look how these radar signals go back to space, and those signals tell us about the surface and the atmosphere.

Satellite technology emerged during the space race for cold war domination.

Go. They need spark plugs. Go.

But in meteorology, at least, observations are now open to all.

A really key turning point was around 1980, when we suddenly had an influx of lots of satellite data.

It's one of the success stories of science, in a sense. US satellites, Chinese satellites, European satellites will all be exchanged across the different weather centres, like ourselves. And everyone needs to share what they have to build the bigger picture.

Back on the surface conventional old school observing and sampling is still very important. Verifying the satellite data from above and recording details that can't be seen from space.

So satellite observations are fantastic when it comes to looking at very large areas of the sea surface. Very limited in their ability to penetrate down into the depths of the ocean. What these pieces of kit allow us to do is to peel back the surface layer and look at some of the most important processes that are going on way out of the reach of satellite observations.

This kind of hands on sampling is helping researchers understand how our seas are changing as they warm to record highs and absorb ever increasing concentrations of carbon dioxide.

Nearly 30 per cent of the carbon dioxide that we've produced has ended up in the ocean. We know that 90 per cent of the heat that has occurred because of global warming has ended up in the ocean. Not only is the ocean the repository for all of these climate impacts, potential solutions exist there.

However, it's really important that we gather data and knowledge to understand how the climate and the ocean interact, so that when we do start to make decisions about how we manage climate change, we make really good decisions and don't end up exacerbating what is already a pretty dangerous situation.

The Plymouth Marine Laboratory and other research organisations are also developing fleets of autonomous self-navigating vessels and submersibles. They are particularly useful to explore deep seas around remote polar regions, which are warming much faster than expected.

We need to care about the polar regions because what happens there doesn't stay there. Sea ice is fundamental in helping drive the circulation patterns. It helps drive the weather conditions that we have in the lower latitudes. It takes nutrients from the surface waters down into the deep ocean and circulates that temperature and those nutrients around the planet, giving life to many other regions of the world.

Climate scientists rely on conventional physics and computer modelling to calculate and predict long term trends. But in shorter term forecasting, AI and machine learning systems are showing great promise.

The data-driven weather forecasting is now learning how to forecast future weather from looking at mind bogglingly large data sets of past weather and starting to work out what are the interactions between temperature structures, humidity structures, and winds.

In one stress test using historical data a machine learning model accurately predicted a kind of strange one-off event that does not follow normal or established patterns.

In 2012, there was a hurricane called Sandy. Instead of turning to the east, it turned to the west and hit New York. It's a very famous event, a once in however many 100 years you want to quote event. And you might think that a machine learning system that learns patterns, normal behaviour it might fail to predict that sort of event, but it didn't.

Nature can still take forecasters by surprise with exceptional events, such as Hurricane Sandy, or more recently, Otis, a predicted tropical storm that developed unexpectedly into a severe and devastating hurricane in October 2023.

But advanced data gathering and processing techniques are recording and accurately predicting more frequent extreme weather across a changing climate, yet the greenhouse gas emissions driving these changes continue to rise.

It does seem to be that we're lagging behind the ambition set out in the Paris Accord. But that said, we're still going to need to manage and adapt to the impacts when they come along.

Having that understanding of how our climate is changing, why it is changing, where it is changing most rapidly is really important.

We have a truly international collaboration. It demonstrates that the science is robust. And there is no better time, no more important time, to continue to observe, monitor, and predict global climate.