Aardvark AI: Revolutionizing Weather Forecasting with Desktop-Powered Predictions

AI can forecast the weather in seconds without needing supercomputers

While earlier weather-forecasting AIs have replaced some tasks done by traditional models, new research uses machine learning to replace the entire process, making it much faster An AI weather program running for a single second on a desktop can match the accuracy of traditional forecasts that take hours or days on powerful supercomputers, claim its creators. Weather forecasting has, since the 1950s, relied on physics-based models that extrapolate from observations made using satellites, balloons and weather stations. But these calculations, known as numerical weather prediction (NWP), are extremely intensive and rely on vast, expensive and energy-hungry supercomputers. In recent years, researchers have tried to streamline this process by applying AI. Google scientists last year created an AI tool that could replace small chunks of complex code in each cell of a weather model, cutting the computer power required dramatically. DeepMind later took this even further and used AI to replace the entire forecast. This approach has been adopted by the European Centre for Medium-Range Weather Forecasts (ECMWF), which launched a tool called the Artificial Intelligence Forecasting System last month. But this gradual expansion of AI's role in weather prediction has fallen short of replacing all traditional number-crunching - something a new model created by Richard Turner at the University of Cambridge and his colleagues seeks to change. Turner says previous work was limited to forecasting, and passed over a step called initialisation, where data from satellites, balloons and weather stations around the world is collated, cleaned, manipulated and merged into an organised grid that the forecast can start from. "That's actually half the computational resources," says Turner. The researchers created a model called Aardvark Weather that, for the first time, replaces both the forecast and initialisation stages. It uses just 10 per cent of the input data that existing systems do, but can achieve results comparable to the latest NWP forecasts, report Turner and his colleagues in a study assessing their method. Generating a full forecast, which would take hours or even days on a powerful supercomputer for an NWP forecast, can be done in approximately 1 second on a single desktop computer using Aardvark. However, Aardvark is using a grid model of Earth's surface with cells that are 1.5 degrees square, while the ECMWF's ERA5 model uses a grid with cells as small as 0.3 degrees. This means Aardvark's model is too coarse to pick up on complex and unexpected weather patterns, says David Schultz at the University of Manchester, UK. "There's a lot of unresolved things going on that could blow up your forecast," says Schultz. "They are not representing the extremes at all. They can't resolve it at this scale." Turner argues that Aardvark can actually beat some existing models in picking up unusual events such as cyclones, and that it produces results comparable to cutting-edge NWP forecasts. But he concedes that AI models like his also rely entirely on those physics-based models for training. "It absolutely doesn't work if you take their training data away and just use the observational data to train off," he says. "We did try to do that, and go completely physics model-free, but that didn't work." He believes the future of weather forecasting may be scientists working on ever-more accurate physics-based models, which are then used to train AI models that replicate their output faster and with less hardware. Some are even more optimistic about the prospects of AI. Nikita Gourianov at the University of Oxford believes that, in time, AI will be able to create weather forecasts that actually surpass NWP. These will be trained on observational and historical weather data alone, creating accurate forecasts entirely independent of NWP, he says. "It's a question of scale, but also a question of cleverness. You have to be clever with how you feed the data in - and how you structure the neural network."

Aardvark beats groundhogs and computer for weather forecasts

PC-size ML prediction model predicted to be as good as a super at fraction of the cost Aardvark, a novel machine learning-based weather prediction system, teases a future where supercomputers are optional for forecasting - but don't pull the plug just yet. Academics affiliated with the Alan Turing Institute in the UK and other institutions claim they have developed a weather prediction model that can be trained and run on a desktop computer at a fraction of the cost and time currently incurred by using supercomputers. Building on work done at the University of Cambridge, Richard Turner, research lead in AI for weather prediction at the Alan Turing Institute and a professor of machine learning at the University of Cambridge, and Scott Hosking, interim director of science and innovation at The Alan Turing Institute, reckon Aardvark's results are good enough that it can replace the entire numerical weather prediction (NWP) pipeline. The NWP pipeline follows a three-step process. First, observational data is gathered from satellites, weather balloons, ground stations, ships, aircraft, and buoys, and combined with a recent forecast to estimate the current state of the atmosphere. Second, this estimate is fed into a complex computational model that simulates atmospheric physics to generate forecasts. Finally, those raw forecasts are post-processed to correct biases, improve local accuracy, and incorporate human forecaster input. Other recent work improving weather forecasting with machine learning, such as Google DeepMind's GenCast model, has focused on step two - the computational model - whereas Aardvark is said to be capable of replacing all three steps. As described in an article published in the science journal Nature this month, "Aardvark provides accurate forecasts that are orders of magnitude quicker to generate than existing systems, without any reliance on NWP products at deployment time. Generating a full forecast from observational data takes approximately one second on four NVIDIA A100 GPUs, compared to the approximately 1,000 node-hours required by HRES to perform data assimilation and forecasting alone, before accounting for downstream local models and processing." The resulting forecast, it's claimed, is as accurate as America's Global Forecast System (GFS). Yet it relies on only about a tenth of the observational data used in traditional NWP systems. "We only used 10 percent of the data as we're quite limited in terms of computational resources in my academic lab compared to, say, technology companies," Turner told The Register. "With more data, I would expect the system to perform better. Aardvark is able to compete using only a fraction of the data partly because the AI models are trained directly to solve the task of interest (forecasting)." Aardvark still requires some refinement as it does not yet have the resolution of Europe's Integrated Forecast System (IFS). That's being worked on, however, and the researchers expect to add various specialized modules to focus on specific types of forecasts like hurricanes, floods, and other extreme weather events. Aardvark consists of three components: An encoder, a processor, and a set of decoder modules. The encoder has about 31 million parameters and takes 13 hours to train. The processor contains about 54 million parameters and requires eight hours of training on ERA5, a weather data set, followed by 3 hours of fine-tuning using the encoder's output. There are eleven decoder modules, each of which has about 2 million parameters and takes 30 minutes to train. Allowing a further two hours for end-to-end fine-tuning, Aardvark can be trained in about 100 GPU hours. Once that's done, Aardvark can create forecasts on a desktop computer within minutes. As a point of comparison, Google claims its GenCast model can produce a 15-day forecast in eight minutes using a single Google Cloud TPU. Relevant source code to replicate the results in the Nature paper is presently restricted, though the plan is to eventually let Aardvark loose. "Everything will be open sourced when the print version of Aardvark goes live and anyone will be able to download it and train the system on tasks of interest," said Turner. ®

Fully AI-driven weather prediction system delivers accurate forecasts faster with less computing power

A new AI weather prediction system, Aardvark Weather, can deliver accurate forecasts tens of times faster and using thousands of times less computing power than current AI and physics-based forecasting systems, according to research published in Nature. Aardvark has been developed by researchers from the University of Cambridge supported by the Alan Turing Institute, Microsoft Research and the European Centre for Medium Range Weather Forecasting, providing a blueprint for a completely new approach to weather forecasting with the potential to transform current practices. The weather forecasts that people rely upon are currently generated through a complex set of stages, each taking several hours to run on bespoke supercomputers. Aside from daily usage, the development, maintenance and deployment of these complex systems requires significant time and large teams of experts. More recently, research by Huawei, Google, and Microsoft has demonstrated that one component of this 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 deployed by the European Centre for Medium Range Weather Forecasts. But with Aardvark, researchers have replaced the entire weather prediction pipeline with a single, simple machine learning model. The 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 that predictions are now achievable 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 and it is also competitive with United States Weather Service forecasts that use input from dozens of weather models and analysis by expert human forecasters. 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 predicting temperatures for African agriculture or wind speeds for a renewable energy company in Europe. This contrasts with traditional weather prediction systems where creating a customized system takes years of work by large teams of researchers. 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. Professor Richard Turner, Lead Researcher for Weather Prediction at the Alan Turing Institute and Professor of Machine Learning in the Department of Engineering at the University of Cambridge, said, "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. "Importantly, 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." Anna Allen, lead author from the University of Cambridge, added, "These results are just the beginning of what Aardvark can achieve. 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." Matthew Chantry, Strategic Lead for Machine Learning at ECMWF, remarked, "We have been thrilled to collaborate on this project, which explores the next generation of weather forecasting systems -- part of our mission to develop and deliver operational AI-weather forecasting while openly sharing data to benefit science and the wider community. It is essential that academia and industry work together to address technological challenges and leverage new opportunities that AI offers. Aardvark's approach combines both modularity with end-to-end forecasting optimization, ensuring effective use of the available datasets." Dr. Chris Bishop, Technical Fellow and Director, Microsoft Research AI for Science, stated, "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." Dr. Scott Hosking, Director of Science and Innovation for Environment and Sustainability at The Alan Turing Institute, observed, "Unleashing AI's potential will transform decision-making for everyone, from policymakers and emergency planners to industries that rely on accurate weather forecasts. Aardvark's breakthrough is not just about speed, it's about access. By shifting weather prediction from supercomputers to desktop computers, we can democratize forecasting, making these powerful technologies available to developing nations and data-sparse regions around the world." Next steps for Aardvark include developing a new team within the Alan Turing Institute led by Professor Richard Turner, exploring the potential to deploy Aardvark in the Global South, and integrating the technology into the Institute's wider work to develop high-precision environmental forecasting for weather, oceans and sea ice.

AI weather forecast project eyes access through desktop computers

Artificial intelligence-powered forecasts as fast and accurate as those generated by the best equipped public weather services could be delivered from desktop computers around the world under an international project unveiled on Thursday. The developers of the Aardvark prediction model hope it will "democratise" the AI-driven weather forecasting revolution by bringing it within reach of countries with fewer resources in Africa and elsewhere. The details were published in Nature. The initiative led by the UK's Alan Turing Institute, with partners including Cambridge university, European Centre for Medium-Range Weather Forecasts and Microsoft, comes amid a flurry of big technical advances in weather forecasting announced by companies such as Google DeepMind and Nvidia as well as public meteorological offices in recent months. "Aardvark's breakthrough is not just about speed, it's about access," said Scott Hosking of the Turing Institute. "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." Researchers in public and private sectors are using machine learning to make predictions from millions of worldwide weather observations rather than crunching the data in supercomputers with conventional physics-based equations. But these new AI systems generally require intensive processing to assimilate the data and set the initial conditions before each run. In contrast, Aardvark is an "end-to-end" model. It dispenses with the compute intensive first step and works directly on observations from satellites, weather stations and other sensors, generating both global and local forecasts. That cuts the energy and processing power required by a factor of many thousands. Although Aardvark is still in its experimental phase, an evaluation by its developers showed that it outperformed the US Global Forecast System on many variables. The team is working to deploy Aardvark in regions poorly served by the global weather models generated by meteorological centres in the industrialised world, particularly in Africa. Aardvark could be transformative in west Africa if adapted to "local contexts" to give more precise information on phenomena such as intense rain systems, said Amadou Gaye, professor of climate physics at Senegal's Université Cheikh Anta Diop Dakar. That would help better predict both damaging flooding episodes and precipitation patterns that affect the harvesting of crops such as peanuts, said Gaye, who has been working with the Turing Institute. He compared the potential impact to the arrival of mobile phones, which had delivered fundamental services such as banking -- and weather forecasts -- to farmers. Meteorological conditions make it easier to forecast the weather a few weeks into the future in Africa than in temperate regions of the northern hemisphere, because the continent's large-scale weather patterns are much more stable, said Richard Turner, project leader at the Turing Institute. "There's a big opportunity here for AI to make 'sub-seasonal' forecasts four to six weeks ahead, which has rather been ignored in the north because it is so hard to do there -- a big win for instance for African agriculture," he added. Although Microsoft was an industrial partner in the project, the company does not plan to commercialise Aardvark, Turner said. "Everything is completely open source." Suzanne Gray, professor of meteorology at Reading university, who is not involved in the project, said: "Aardvark's demonstrated capability to produce both global and local station forecasts directly from observations . . . is hugely impressive and showcases the massive potential of machine learning in this weather forecasting." But she said more development was needed for Aardvark to generate all the variables and fine spatial resolution needed for the forecasts and weather warnings issued routinely by public bodies such as the UK Met Office.

New AI is better at weather prediction than supercomputers -- and it consumes 1000s of times less energy

Aardvark Weather generates forecasts more quickly and with less computing power than existing forecasting systems. (Image credit: BroadcastNews via Shutterstock) A new artificial intelligence (AI)-driven weather prediction system could transform forecasting, researchers predict The system, dubbed Aardvark Weather, generates forecasts tens of times faster than traditional forecasting systems using a fraction of the computing power, researchers reported Thursday (March 20) in the journal Nature. "The 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," Richard Turner, an engineer at the University of Cambridge in the United Kingdom, said in a statement. Current weather forecasts are generated by inputting data into complex physics models, a multi-stage process that requires several hours on a dedicated supercomputer. Aardvark Weather circumvents this demanding process: the machine learning model uses raw data from satellites, weather stations, ships and weather balloons to make its predictions without relying on atmospheric models. Satellite data are particularly important for the model's predictions, the team noted. Related: Google builds an AI model that can predict future weather catastrophes This new approach could offer major advantages in terms of cost, speed and accuracy of weather forecasts, the researchers claimed. Instead of requiring a supercomputer and a dedicated team, Aardvark Weather can generate a forecast on a desktop computer in just a few minutes. The team compared Aardvark's performance to existing forecasting systems that generate global predictions. Using just 8% of the observational data that traditional forecasting systems need, Aardvark outperformed the U.S. national Global Forecast System (GFS)system and was comparable to forecasts made by the United States Weather Service. However, Aardvark's spatial resolution is somewhat lower than those of current forecasting systems, which could make its initial predictions less relevant for hyper-local weather forecasting. Aardvark Weather operates at 1.5-degree resolution, meaning each box in its grid covers 1.5 degrees of latitude and 1.5 degrees of longitude. For comparison, the GFS uses a 0.25-degree grid. However, the researchers also said that because the AI learns from the data it is fed, it could be tailored to predict weather in specific arenas -- such as temperatures for African agriculture or wind speeds for renewable energy in Europe. Aardvark can incorporate higher-resolution regional data, where they exist, to refine local forecasts. "These results are just the beginning of what Aardvark can achieve," study coauthor Anna Allen, of the University of Cambridge, said in the statement. "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." Aardvark could also support forecasting centers in areas of the world that lack the resources to refine global forecasts into high-resolution regional predictions, the researchers said. "Aardvark's breakthrough is not just about speed, it's about access," Scott Hosking, an AI researcher at The Alan Turing Institute in the U.K., said in the statement. "By shifting weather prediction from supercomputers to desktop computers, we can democratize forecasting, making these powerful technologies available to developing nations and data-sparse regions around the world."

Cambridge researchers unveil faster and more accurate AI weather system that rivals supercomputers

In a nutshell: Aardvark Weather, an AI-based system, promises to significantly enhance weather forecasting by delivering predictions dozens of times faster while using thousands of times less computing power than current methods. This system has been developed by researchers at the University of Cambridge, with support from the Alan Turing Institute, Microsoft Research, and the European Centre for Medium Range Weather Forecasts. The speed and efficiency of modern forecasting systems are vital, as traditional methods rely on powerful supercomputers and extensive teams of experts, often requiring several hours to produce forecasts. Recent innovations from tech giants such as Huawei, Google, and Microsoft have demonstrated that AI can significantly improve specific aspects of the forecasting process, including numerical solvers, which are crucial in weather forecasting as they simulate how atmospheric conditions evolve over time. These companies have achieved faster and more accurate predictions by integrating AI into these solvers. As one example, Google has been developing AI models for weather forecasting and is currently marketing two models to its enterprise cloud customers. Developed by Google DeepMind, the models use historical weather data to predict future conditions 10 to 15 days in advance. Aardvark represents a significant advancement by replacing traditional forecasting processes with a single, streamlined machine-learning model. Using a standard desktop computer, it can process data from various sources, including satellites and weather stations, to generate global and local forecasts in minutes. "Aardvark reimagines current weather prediction methods, offering the potential to make weather forecasts faster, cheaper, more flexible, and more accurate than ever before," explained 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." Despite operating with only a fraction of the data used by existing systems, Aardvark surpasses the U.S. national GFS forecasting system in several key metrics and remains competitive with forecasts from the National Weather Service, which typically involve multiple models and expert analysis. "These results are just the beginning of what Aardvark can achieve," noted first author Anna Allen from Cambridge's Department of Computer Science and Technology. She said the end-to-end learning approach can be easily applied to other weather forecasting problems, such as hurricanes, wildfires, and tornadoes. It can also be used for broader Earth system forecasting, including air quality, ocean dynamics, and sea ice prediction. One of the most interesting 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 predicting temperatures to support African agriculture or wind conditions for European renewable energy firms. This contrasts sharply with traditional systems, which require years of work by large teams to customize. This capability has the potential to transform weather prediction in developing countries, where access to expertise and computational resources is limited. "By shifting weather prediction from supercomputers to desktop computers, we can democratize forecasting, making these powerful technologies available to developing nations and data-sparse regions around the world," said Dr. Scott Hosking from The Alan Turing Institute. Aardvark is expected to play a significant role in expanding the scope of weather forecasting. Turner mentioned that the model could eventually accurately predict eight-day forecasts, surpassing the capabilities of current models by three days. This advancement, along with Aardvark's adaptability and efficiency, positions it as a transformative force in meteorology. The next steps for Aardvark include developing a new team within the Alan Turing Institute that will explore deploying the technology in the global south and integrating it into broader environmental forecasting initiatives.

AI prediction model is a major breakthrough in weather forecasting - Earth.com

New technology is transforming weather prediction. Aardvark Weather is a system that uses artificial intelligence to deliver accurate forecasts in just minutes - right from a regular desktop computer. It's tens of times faster than current methods and requires only a fraction of the computing power. The technology was developed by scientists at the University of Cambridge, with support from the Alan Turing Institute, Microsoft Research, and the European Centre for Medium-Range Weather Forecasts (ECMWF). The system lays the foundation for a new kind of forecasting that could shift the way we understand and prepare for weather around the world. Right now, most weather forecasts come from massive systems that require supercomputers to run. These systems are highly accurate but expensive, complex, and slow. Each step in the forecasting process - from collecting data to generating local predictions - takes hours and involves many technical stages. Even minor updates to these systems take large teams of experts and years to implement. Recently, companies like Huawei, Google, and Microsoft have demonstrated that AI can replace one of those steps - the numerical solver, which calculates how weather systems evolve. This has made forecasting both faster and more accurate. ECMWF is already using a mix of these traditional and AI-based tools. Aardvark goes a step further. It replaces the entire pipeline with a single machine learning model. The system takes real-time data from satellites, sensors, and weather stations, and immediately outputs both local and global forecasts. The result? Lightning-fast predictions that can be generated on a desktop computer. And because it's trained directly on data, the system avoids many of the hurdles that come with designing new forecasting models from scratch. Professor Richard Turner is the lead researcher for Weather Prediction at the Alan Turing Institute and Professor of Machine Learning in the Department of Engineering at the University of Cambridge. "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 Turner. "Importantly, 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." Even when Aardvark uses only 10% of the data that conventional models need, it still outperforms the United States' national GFS forecasting system in many categories. Aardvark also holds its own against forecasts produced by expert analysts working with dozens of models. This efficiency makes it not only fast, but incredibly versatile. The model can be quickly trained to produce location-specific predictions - like rainfall for African farms or wind speeds for European wind farms. That level of customization used to take years to build. With Aardvark, it can happen in weeks. "These results are just the beginning of what Aardvark can achieve," said Anna Allen, lead author from the University of Cambridge. "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." One of Aardvark's most promising features is accessibility. Since it doesn't rely on expensive supercomputers, the model can be used anywhere - even in countries with limited computing infrastructure. This levels the playing field for weather prediction and planning. "We have been thrilled to collaborate on this project which explores the next generation of weather forecasting systems - part of our mission to develop and deliver operational AI-weather forecasting while openly sharing data to benefit science and the wider community," said Matthew Chantry, Strategic Lead for Machine Learning at ECMWF. "It is essential that academia and industry work together to address technological challenges and leverage new opportunities that AI offers. Aardvark's approach combines both modularity with end-to-end forecasting optimisation, ensuring effective use of the available datasets." Dr. Chris Bishop is the director of Microsoft Research AI for Science. He noted that 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. The next step is to build a team within the Alan Turing Institute focused on expanding Aardvark's use, especially in developing countries. The experts also plan to integrate the model into wider efforts around environmental prediction, including sea ice and ocean behavior. "Unleashing AI's potential will transform decision-making for everyone from policymakers and emergency planners to industries that rely on accurate weather forecasts. Aardvark's breakthrough is not just about speed, it's about access," said Dr. Scott Hosking, Director of Science and Innovation for Environment and Sustainability at The Alan Turing Institute. "By shifting weather prediction from supercomputers to desktop computers, we can democratize forecasting, making these powerful technologies available to developing nations and data-sparse regions around the world." Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.

New AI weather forecasting model outperforms competitors

Why it matters: Applying artificial intelligence to weather prediction holds the promise of significantly advancing forecast precision, reliability and delivery to the developing world. Driving the news: The new model is the result of an international effort among the University of Cambridge, Alan Turing Institute, Microsoft Research and European Centre for Medium-Range Weather Forecasts (ECMWF). Zoom in: The new model, detailed in a study in the journal Nature, is known as Aardvark Weather. It offers what its creators call an "end-to-end AI forecasting system." Yes, but: The AI models developed to date are still somewhat dependent on the work of traditional numerical systems at the initial step of incorporating vast amounts of weather data. Aardvark also uses far fewer observations as inputs compared to both traditional models in use and other AI-driven ones. The intrigue: The researchers tout Aardvark's ability to result in specially-tailored forecasts while being run on a desktop computer, providing results that are available within minutes. Currently, such hyper-focused models can take many months to years to develop and require supercomputers to run. Between the lines: The new, experimental model doesn't eliminate the need for real-world weather data gathering, conventional modeling or human forecasters. While ECMWF has been at the forefront of developing and implementing AI models, NOAA is only beginning to travel down this road in the U.S., with the American private sector moving faster to capitalize on new technologies. What they're saying: "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 Richard Turner, a study coauthor and researcher at the Alan Turing Institute and Cambridge University, in a statement. What we're watching: How the broader field of AI weather modeling evolves and is incorporated into the work of government forecast agencies. Exclusive: New Nvidia model could bolster severe weather forecasts

AI-driven weather prediction breakthrough reported

Researchers say Aardvark Weather uses thousands of times less computing power and is much faster than current systems A single researcher with a desktop computer will be able to deliver accurate weather forecasts using a new AI weather prediction approach that is tens of times faster and uses thousands of times less computing power than conventional systems. Weather forecasts are currently generated through a complex set of stages, each taking several hours to run on bespoke supercomputers, requiring large teams of experts to develop, maintain and deploy them. Aardvark Weather provides a blueprint to replace the entire process by training an AI on raw data from weather stations, satellites, weather balloons, ships and planes from around the world to enable it to make predictions. This offers the potential for vast improvements in forecast speed, accuracy and cost, according to research published on Thursday in Nature from the University of Cambridge, the Alan Turing Institute, Microsoft Research and the European Centre for Medium-Range Weather Forecasts (ECMWF). Richard Turner, a professor of machine learning at the University of Cambridge, said the approach could be used to quickly provide bespoke forecasts for specific industries or locations, for example 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, while supercomputers take hours to process measurements from the real world in order to build forecasting models. "This is a completely different approach to what people have done before. The writing's on the wall that this is going to transform things, it's going to be the new way of doing forecasting," Turner said. He said the model would eventually be able to produce accurate eight-day forecasts, compared with five-day forecast at present, as well as hyper-localised predictions. Dr Scott Hosking, the director of science and innovation for environment and sustainability at the Alan Turing Institute, said the breakthrough could "democratise forecasting" by making powerful technologies available to developing nations around the world, as well as assisting policymakers, emergency planners and industries that rely on accurate weather forecasts. Dr Anna Allen, the lead author of the paper, from the University of Cambridge, noted that the findings paved the way for better forecasts of natural disasters such as hurricanes, wildfires and tornadoes, as well as other climatic issues such as air quality, ocean dynamics and sea ice predictions. Aardvark builds on recent research by Huawei, Google, and Microsoft demonstrating that one step of the weather prediction process known as the numerical solver, which calculates how weather evolves over time, can be replaced with AI to produce faster and more accurate predictions. This approach is already being deployed by the ECMWF. The researchers said that using just 10% of the input data that existing systems required, Aardvark could already outperform the US national GFS forecasting system in certain respects, and was competitive with United States Weather Service forecasts.

New AI Model Promises Faster, Smarter Weather Predictions - Decrypt

Aardvark Weather, a new AI model developed by researchers in the UK and Canada, could mark a turning point in global weather forecasting by replacing traditional weather simulations with artificial intelligence to maximize cost efficiency and accuracy. Researchers from the University of Cambridge, the Vector Institute at the University of Toronto, and the Alan Turing Institute unveiled the new findings in a recent report published in Nature. Unlike conventional forecasting tools that simulate atmospheric physics through complex equations, Aardvark Weather is a "deep learning" model that generates global forecasts for wind, humidity, geopotential, and temperature at multiple pressure levels. It also delivers local station forecasts for 2-meter temperature and 10-meter wind speed. Deep learning is a subset of machine learning that teaches computers to recognize patterns in large amounts of data. "At the moment, there are some computationally expensive components in the forecasting pipeline," postdoctoral fellow at the University of Toronto's Vector Institute James Requeima told Decrypt. "We've been able to replace many of these time-consuming parts with much lighter-weight models trained to perform the same tasks." By making those components more efficient, Aardvark could run forecasts more often and at higher resolutions, improving speed and accuracy. As Requeima explained, the team designed components to replace each step in the forecasting pipeline, which involves turning raw observational data into a weather forecast. "We found that once these machine learning components are chained together, the overall performance improves significantly," he said. "By fine-tuning the entire pipeline for the final task we're targeting, we can optimize each component not just for its isolated role, but for how it contributes to the outcome we care most about." The project also included researchers from Microsoft Research Cambridge, the European Centre for Medium-Range Weather Forecasts (ECMWF), and the British Antarctic Survey. Aardvark Weather uses raw atmospheric data -- like pressure, temperature, and relative humidity measurements -- to produce high-resolution global and local forecasts. The system is built around three neural components: an encoder, a processor, and a decoder. To improve Aardvark's performance and accuracy, components are first pre-trained on ERA5 reanalysis data -- a high-quality historical dataset from ECMWF -- and then fine-tuned using real-world weather observations. "Data assimilation, in general, works like an autoregressive procedure. You start with the current atmospheric forecast, generated by large dynamical systems that estimate its present state. At time zero, you have this initial state," Requeima said. "But data assimilation also needs to incorporate real-time measurements from remote sensors. So, you gather actual observations alongside the model's forecast and adjust your atmosphere estimate accordingly." According to the report, Aardvark can generate a full global forecast using four NVIDIA A100 GPUs in just one second compared to the hours needed by older models like the European Centre for Medium-Range Weather Forecasts' high-resolution forecast. This drastic reduction in computing requirements makes high-quality, customizable forecasting accessible to regions and agencies without the resources to operate full-scale NWP systems. It also enables much faster fine-tuning of the model. Aardvark joins a growing suite of tools aimed at helping meteorologists predict and respond to extreme weather events. During recent storms, such as Hurricanes Helene and Milton, which battered the U.S. East Coast in October 2024, forecasters emphasized the importance of AI in improving storm intensity prediction. Looking ahead, Requeima noted that the team plans to open source Aardvark to make the technology more widely accessible. "I think it's an important step toward democratizing weather modeling -- making it more lightweight and accessible to the public," he said. "That's our hope. It also represents a major advancement in end-to-end weather modeling, particularly through a data-driven, machine learning approach."

This AI Weather Forecasting System Does Not Require Supercomputers

It is said to have outperformed the US national GFS forecasting system Aardvark Weather, an artificial intelligence (AI) system for weather forecasting that offers several advantages over traditional methods, was unveiled on Thursday. The fully AI-driven system is said to be capable enough to replace the existing systems that require custom supercomputers and a team of experts to deliver predictions. It is also said to improve prediction accuracy and speed, and provide hyperlocal forecasts. A large group of researchers from academic institutions and tech companies worked together to develop the AI system. The research paper was published in the Nature journal (can be accessed here), and details how a single, unified AI model replaces supercomputers and expert supervision. Notably, last year Google unveiled its GenCast model which uses AI to replace the numerical solver to improve the accuracy and speed of weather predictions. However, the scale of Aardvark Weather is much larger as it aims to replace the entire forecasting system. The AI system was developed by researchers from the University of Cambridge, Microsoft Research AI for Science, Google DeepMind, the University of Toronto, The Alan Turing Institute, and the European Centre for Medium-Range Weather Forecasts (ECMWF). Aardvark Weather collects data from weather satellites, weather balloons, stations, and other sensors, and it can process and analyse the data to produce both global and local forecasts. What typically takes custom supercomputers and weather science experts a few hours can be generated in a few minutes with the AI system. Researchers claimed that the AI system was able to outperform the US national Global Forecast System (GFS) on multiple variables with just 10 percent of the input data. It is also said to perform on par with the US Weather Service forecasts. Aardvark can also be used to provide bespoke weather predictions for specific industries or locations. "These results are just the beginning of what Aardvark can achieve. 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," said first author Anna Allen, from Cambridge's Department of Computer Science and Technology, in a statement.