̽��ֱ�� of Cambridge - Richard Turner

Opinion: AI can democratise weather forecasting

lw355 — Tue, 01 Apr 2025 14:12:44 +0000

AI will give us the next leap forward in forecasting the weather, says Richard Turner, and make it available to all countries, not just those with access to high-quality data and computing resources.

Fully AI driven weather prediction system could start revolution in forecasting

sc604 — Thu, 20 Mar 2025 15:56:54 +0000

̽��ֱ��system, Aardvark Weather, has been supported by the Alan Turing Institute, Microsoft Research and the European Centre for Medium Range Weather Forecasts. It provides a blueprint for a new approach to weather forecasting with the potential to transform current practices. The results are reported in the journal Nature.

“Aardvark reimagines current weather prediction methods offering the potential to make weather forecasts faster, cheaper, more flexible and more accurate than ever before, helping to transform weather prediction in both developed and developing countries,” said Professor Richard Turner from Cambridge’s Department of Engineering, who led the research. “Aardvark is thousands of times faster than all previous weather forecasting methods.”

Current weather forecasts are generated through a complex set of stages, each taking several hours to run on powerful supercomputers. Aside from daily usage, the development, maintenance and use of these systems require significant time and large teams of experts.

More recently, research by Huawei, Google, and Microsoft has shown that one component of the weather forecasting pipeline, the numerical solver (which calculates how weather evolves over time), can be replaced with AI, resulting in faster and more accurate predictions. This combination of AI and traditional approaches is now being used by the European Centre for Medium Range Weather Forecasts (ECMWF).

But with Aardvark, researchers have replaced the entire weather prediction pipeline with a single, simple machine learning model. ̽��ֱ��new model takes in observations from satellites, weather stations and other sensors and outputs both global and local forecasts.

This fully AI driven approach means predictions that were once produced using many models – each requiring a supercomputer and a large support team to run – can now be produced in minutes on a desktop computer.

When using just 10% of the input data of existing systems, Aardvark already outperforms the United States national GFS forecasting system on many variables. It is also competitive with United States Weather Service forecasts that use input from dozens of weather models and analysis by expert human forecasters.

“These results are just the beginning of what Aardvark can achieve,” said first author Anna Allen, from Cambridge’s Department of Computer Science and Technology. “This end-to-end learning approach can be easily applied to other weather forecasting problems, for example hurricanes, wildfires, and tornadoes. Beyond weather, its applications extend to broader Earth system forecasting, including air quality, ocean dynamics, and sea ice prediction.”

̽��ֱ��researchers say that one of the most exciting aspects of Aardvark is its flexibility and simple design. Because it learns directly from data it can be quickly adapted to produce bespoke forecasts for specific industries or locations, whether that's predicting temperatures for African agriculture or wind speeds for a renewable energy company in Europe.

This contrasts to traditional weather prediction systems where creating a customised system takes years of work by large teams of researchers.

“ ̽��ֱ��weather forecasting systems we all rely on have been developed over decades, but in just 18 months, we’ve been able to build something that’s competitive with the best of these systems, using just a tenth of the data on a desktop computer,” said Turner, who is also Lead Researcher for Weather Prediction at the Alan Turing Institute.

This capability has the potential to transform weather prediction in developing countries where access to the expertise and computational resources required to develop conventional systems is not typically available.

“Unleashing AI’s potential will transform decision-making for everyone from policymakers and emergency planners to industries that rely on accurate weather forecasts,” said Dr Scott Hosking from ̽��ֱ��Alan Turing Institute. “Aardvark’s breakthrough is not just about speed, it’s about access. By shifting weather prediction from supercomputers to desktop computers, we can democratise forecasting, making these powerful technologies available to developing nations and data-sparse regions around the world.”

“Aardvark would not have been possible without decades of physical-model development by the community, and we are particularly indebted to ECMWF for their ERA5 dataset which is essential for training Aardvark,” said Turner.

“It is essential that academia and industry work together to address technological challenges and leverage new opportunities that AI offers,” said Matthew Chantry from ECMWF. “Aardvark’s approach combines both modularity with end-to-end forecasting optimisation, ensuring effective use of the available datasets."

“Aardvark represents not only an important achievement in AI weather prediction but it also reflects the power of collaboration and bringing the research community together to improve and apply AI technology in meaningful ways,” said Dr Chris Bishop, from Microsoft Research.

̽��ֱ��next steps for Aardvark include developing a new team within the Alan Turing Institute led by Turner, who will explore the potential to deploy Aardvark in the global south and integrate the technology into the Institute’s wider work to develop high-precision environmental forecasting for weather, oceans and sea ice.

Reference:
Anna Allen, Stratis Markou et al. ‘End-to-end data-driven weather prediction.’ Nature (2025). DOI: 10.1038/s41586-025-08897-0

Adapted from a media release by ̽��ֱ��Alan Turing Institute

A new AI weather prediction system, developed by researchers from the ̽��ֱ�� of Cambridge, can deliver accurate forecasts tens of times faster and using thousands of times less computing power than current AI and physics-based forecasting systems.

̽��ֱ��Alan Turing Institute

Professor Richard Turner using Aardvark Weather

̽��ֱ��text in this work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. Images, including our videos, are Copyright © ̽��ֱ�� of Cambridge and licensors/contributors as identified. All rights reserved. We make our image and video content available in a number of ways – on our main website under its Terms and conditions, and on a range of channels including social media that permit your use and sharing of our content under their respective Terms.

Yes

Architecting the future

skbf2 — Tue, 08 Dec 2020 16:17:50 +0000

Arm is working with Cambridge researchers to make our phones and computers more secure, more efficient and ready for the digital revolution.

̽��ֱ��uncertain unicycle that taught itself and how it’s helping AI make good decisions

lw355 — Wed, 14 Feb 2018 13:51:19 +0000

In the centre of the screen is a tiny unicycle. ̽��ֱ��animation starts, the unicycle lurches forward and falls. This is trial #1. It’s now trial #11 and there’s a change – an almost imperceptible delay in the fall, perhaps an attempt to right itself before the inevitable crash. “It’s learning from experience,” nods Professor Carl Edward Rasmussen.

After a minute, the unicycle is gently rocking back and forth as it circles on the spot. It’s figured out how this extremely unstable system works and has mastered its goal. “ ̽��ֱ��unicycle starts with knowing nothing about what’s going on – it’s only been told that its goal is to stay in the centre in an upright fashion. As it starts falling forwards and backwards, it starts to learn,” explains Rasmussen, who leads the Computational and Biological Learning Lab in the Department of Engineering. “We had a real unicycle robot but it was actually quite dangerous – it was strong – and so now we use data from the real one to run simulations, and we have a mini version.”

Rasmussen uses the self-taught unicycle to demonstrate how a machine can start with very little data and learn dynamically, improving its knowledge every time it receives new information from its environment. ̽��ֱ��consequences of adjusting its motorised momentum and balance help the unicycle to learn which moves were important in helping it to stay upright in the centre.

“This is just like a human would learn,” explains Professor Zoubin Ghahramani, who leads the Machine Learning Group in the Department of Engineering. “We don’t start knowing everything. We learn things incrementally, from only a few examples, and we know when we are not yet confident in our understanding.”

Ghahramani’s team is pioneering a branch of AI called continual machine learning. He explains that many of the current forms of machine learning are based on neural networks and deep learning models that use complex algorithms to find patterns in vast datasets. Common applications include translating phrases into different languages, recognising people and objects in images, and detecting unusual spending on credit cards.

“These systems need to be trained on millions of labelled examples, which takes time and a lot of computer memory,” he explains. “And they have flaws. When you test them outside of the data they were trained on they tend to perform poorly. Driverless cars, for instance, may be trained on a huge dataset of images but they might not be able to generalise to foggy conditions.

“Worse than that, the current deep learning systems can sometimes give us confidently wrong answers, and provide limited insight into why they have come to particular decisions. This is what bothers me. It’s okay to be wrong but it’s not okay to be confidently wrong.”

̽��ֱ��key is how you deal with uncertainty – the uncertainty of messy and missing data, and the uncertainty of predicting what might happen next. “Uncertainty is not a good thing – it’s something you fight, but you can’t fight it by ignoring it,” says Rasmussen. “We are interested in representing the uncertainty.”

It turns out that there’s a mathematical theory that tells you what to do. It was first described by 18th-century English statistician Thomas Bayes. Ghahramani’s group was one of the earliest adopters in AI of Bayesian probability theory, which describes how the probability of an event occurring (such as staying upright in the centre) is updated as more evidence (such as the decision the unicycle last took before falling over) becomes available.

Dr Richard Turner explains how Bayes’ rule handles continual learning: “the system takes its prior knowledge, weights it by how accurate it thinks that knowledge is, then combines it with new evidence that is also weighted by its accuracy.

“This is much more data-efficient than the way a standard neural network works,” he adds. “New information can cause a neural network to forget everything it learned previously – called catastrophic forgetting – meaning it needs to look at all of its labelled examples all over again, like relearning the rules and glossary of a language every time you learn a new word.

“Our system doesn’t need to revisit all the data it’s seen before – just like humans don’t remember all past experiences; instead we learn a summary and we update it as things go on.” Ghahramani adds: “ ̽��ֱ��great thing about Bayesian machine learning is the system makes decisions based on evidence – it’s sometimes thought of as ‘automating the scientific method’ – and because it’s based on probability, it can tell us when it’s outside its comfort zone.”

Ghahramani is also Chief Scientist at Uber. He sees a future where machines are continually learning not just individually but as part of a group. “Whether it’s companies like Uber optimising supply and demand, or autonomous vehicles alerting each other to what’s ahead on the road, or robots working together to lift a heavy load – cooperation, and sometimes competition, in AI will help solve problems across a huge range of industries.”

One of the really exciting frontiers is being able to model probable outcomes in the future, as Turner describes. “ ̽��ֱ��role of uncertainty becomes very clear when we start to talk about forecasting future problems such as climate change.”

Turner is working with climate scientists Dr Emily Shuckburgh and Dr Scott Hosking at the British Antarctic Survey to ask whether machine learning techniques can improve understanding of climate change risks in the future.

“We need to quantify the future risk and impacts of extreme weather at a local scale to inform policy responses to climate change,” explains Shuckburgh. “ ̽��ֱ��traditional computer simulations of the climate give us a good understanding of the average climate conditions. What we are aiming to do with this work is to combine that knowledge with observational data from satellites and other sources to get a better handle on, for example, the risk of low-probability but high-impact weather events.”

“It’s actually a fascinating machine learning challenge,” says Turner, who is helping to identify which area of climate modelling is most amenable to using Bayesian probability. “ ̽��ֱ��data are extremely complex, and sometimes missing and unlabelled. ̽��ֱ��uncertainties are rife.” One significant element of uncertainty is the fact that the predictions are based on our future reduction of emissions, the extent of which is as yet unknown.

“An interesting part of this for policy makers, aside from the forecasting value, is that you can imagine having a machine that continually learns from the consequences of mitigation strategies such as reducing emissions – or the lack of them – and adjusts its predictions accordingly,” adds Turner.

What he is describing is a machine that – like the unicycle – feeds on uncertainty, learns continuously from the real world, and assesses and then reassesses all possible outcomes. When it comes to climate, however, it’s also a machine of all possible futures.

Inset image: read more about our AI research in the ̽��ֱ��'s research magazine; download a pdf; view on Issuu.

Cambridge researchers are pioneering a form of machine learning that starts with only a little prior knowledge and continually learns from the world around it.

This is just like a human would learn. We don’t start knowing everything. We learn things incrementally, from only a few examples, and we know when we are not yet confident in our understanding

Zoubin Ghahramani

̽��ֱ��District

̽��ֱ��text in this work is licensed under a Creative Commons Attribution 4.0 International License. For image use please see separate credits above.

Yes

Noises off: the machine that rubs out noise

lw355 — Wed, 02 Oct 2013 16:20:50 +0000

A noisy restaurant, a busy road, a windy day – all situations that can be intensely frustrating for the hearing impaired when trying to pick out speech in a noisy environment. Some 10 million people in the UK suffer from hearing difficulties and, as helpful as hearing aids are, those who wear them often complain that background noise continues to be a problem.

What if hearing device wearers could choose to filter out all the troublesome sounds and focus on the voices they want to hear? Engineer Dr Richard Turner believes that this is fast becoming a possibility. He is developing a system that identifies the corrupting noise and “rubs it out”.

“ ̽��ֱ��poor performance of current hearing devices in noise is a major reason why six million people in the UK who would benefit from a hearing aid do not use them,” he said. Moreover, as the population ages, a greater number of people will be hindered by the inability to hear clearly. In addition, patients fitted with cochlear implants – devices implanted into the brain to help those whose auditory hair cells have died – suffer from similar limitations.

̽��ֱ��solution lies in the statistics of sound, as Turner explained: “Many interfering noises are immediately recognisable. Raindrops patter on a surface, a fire crackles, talkers babble at a party and the wind howls. But what makes these so-called auditory textures sound the way they do? No two rain sounds are identical because the precise arrangement of falling water droplets is never repeated. Nonetheless, there must be a statistical similarity in the sounds compared with say the crackle of a fire.

“For this reason, we think the brain groups together different aspects of sounds using prior experience of their characteristic statistical structure. We can model this mathematically using a form of statistical reasoning called Bayesian inference and then develop computer algorithms that mimic what the brain is doing.”

̽��ֱ��mathematical system that he and colleagues have developed is capable of being “trained” – a process that uses new methods from the field of machine learning – so that it can recognise sounds. “Rather surprisingly, it seems that a relatively small set of statistics is sufficient to describe a large number of sounds.”

Crucially, the system is capable of telling the difference between speech and audio textures. “What we can now do in an adaptive way is to remove background noise and pass these cleaned up sounds to a listener to improve their perception in a difficult environment,” said Turner, who is working with hearing experts Professor Brian Moore at the Department of Experimental Psychology and Dr Robert Carlyon at the Medical Research Council Cognition and Brain Sciences Unit, with funding from the Engineering and Physical Sciences Research Council.

̽��ֱ��idea is that future devices will have several different modes in which they can operate. These might include a mode for travelling in a car or on a train, a mode for environments like a party or a noisy restaurant, a mode for outdoor environments that are windy, and so on. ̽��ֱ��device might intelligently select an appropriate mode based on the characteristics of the incoming sound. Alternatively, the user could override this and select a processing mode based upon what sorts of noise they wish to erase.

“In a sense we are developing the technology to underpin intelligent hearing devices,” he added. “One possibility would be for users to control their device using an interface on a mobile phone through wireless communication. This would allow users to guide the processing as they wish.”

Turner anticipates a further two years of simulating the effect of modifications that clean up sound before they start to work with device specialists. “If these preliminary tests go well, then we’ll be looking to work with hearing device companies to try to adapt their processing to incorporate these machine learning techniques. If all goes well, we would hope that this technology will be available in consumer devices within 10 years.”

Tinnitus sufferers could also benefit from the technology. Plagued by a constant ringing in the ears, people with tinnitus sometimes use environmental sound generators as a distraction. Such generators offer a limited selection of sounds – a babbling brook, waves lapping, leaves rustling – but, with the new technology, “patients could traverse the entire space of audio textures and figure out where in this enormous spectrum is the best sound for relieving their tinnitus,” added Turner.

̽��ֱ��technology not only holds promise for helping the hearing impaired, but it also has the potential to improve mobile phone communication – anyone who has ever tried to hold a conversation with someone phoning from a crowded room will recognise the possible benefits of such a facility.

Moreover, with 100 hours of video now being uploaded to YouTube every minute, Google has recognised the potential for systems that can recognise audio content and is funding part of Turner’s research. “As an example, a YouTube video containing a conversation that takes place by a busyroadside on a windy day could be automatically categorised based on the speech, traffic and wind noises present in the soundtrack, allowing users to search videos for these categories. In addition, the soundtrack could also be made more intelligible by isolating the speech from the noises – one can imagine users being offered the chance to de-noise their video during the upload process.

“We think this new framework will form a foundation of the emerging field of ‘machine hearing’. In the future, machine hearing will be standard in a vast range of applications from hearing devices, which is a market worth £18 billion per annum, to audio searching, and from music processing tasks to augmented reality systems. We believe this research project will kick-start this proliferation.”

For more information, please contact Louise Walsh (lw355@admin.cam.ac.uk).

Inset image: Dr Richard Turner

Future hearing aids could be adjusted by the wearer to remove background noise using new technology that could also be used to clean up and search YouTube videos.

We are developing the technology to underpin intelligent hearing devices

Richard Turner

̽��ֱ��Machine that Rubs Out Noise

̽��ֱ��District

This work is licensed under a Creative Commons Licence. If you use this content on your site please link back to this page.

Yes