Since the Met Office’s creation over 170 years ago, cutting-edge science and technology has been at the core of what we do. Now, artificial intelligence and machine learning represent the next frontier for meteorology and climate science.
As noted in the Met Office’s Data Science Framework, the Met Office is committed to working with partners across the national and international community to realise the value of data science to the weather and climate endeavour, and stay abreast of all the developments and opportunities associated with this fast-evolving technology.
This partnership includes work on new machine-learning models (for example the FastNet weather prediction model co developed with The Alan Turing Institute) but also helping external partners understand what forecasters need from weather models to accelerate innovation across the industry.
As part of this endeavour, the Met Office is working with Google’s latest machine learning models, including WeatherNext 3, to explore practical usability for operational forecasters within experimental research workflows.
Augmenting physics with computational speed
Existing weather forecasting uses numerical weather prediction, simulating the atmosphere by calculating complex physics-based laws on high-performance computers. This physics-based approach remains the foundation of global meteorology and underpins the Met Office’s forecast, as well as the National Severe Weather Warning Service.
AI-based models approach the challenge of forecasting from a different angle. Trained on extensive historical records and observational datasets, they can generate global multi-day forecast scenarios in minutes, learning patterns in data rather than the physics of the atmosphere. While fast and computationally light at the forecasting stage, AI is not viewed as a replacement for physics-based models, but rather as a complement.
Pairing the physical consistency of existing systems with the rapid scenario generation of machine learning could enable meteorologists to explore broader probabilities, track fast-moving weather systems, and gain earlier indicators of potential high-impact events.
New partnership with Google
Following earlier collaborative research on tropical cyclone prediction, the Met Office and Google signed a new multi-year agreement under which Met Office scientists will be assessing WeatherNext 3 across experimental sandboxes. WeatherNext 3 introduces several technical developments designed for operational forecasting needs, including direct ingestion of live satellite and observational feeds to provide rapid updates on changing conditions.
The WeatherNext 3 model also delivers high-resolution spatial outputs that could allow forecasters to better analyse localised terrain and coastal interfaces across the UK, alongside advanced probabilistic techniques to identify early signals of intense convective rainfall and support renewable energy partners with refined surface wind and solar guidance.
Met Office Science Strategic Head Dr Caroline Bain said: “Our core priority is providing trusted weather intelligence that keeps the public safe. AI models like WeatherNext 3 could add value alongside our physical prediction operational systems, offering our meteorologists new ways to explore atmospheric uncertainty, ensuring innovative science pulls through to impact and directly strengthens dependable public advice.”
Peter Battalaglia, Senior Director, Sustainability research, Google DeepMind said: “National meteorological services set the benchmark for atmospheric science and public safety. Evaluating WeatherNext 3 with the Met Office ensures cutting-edge AI remains grounded in real operational needs, supporting national resilience and early warnings.”
Scientific integrity and the human forecaster
Public trust rests on scientific reliability. Newly emerging AI techniques undergo existing strict validation processes before being used in operational forecasts. Expert human meteorologists remain indispensable: they interpret subtle atmospheric signals, resolve model divergence and deliver the clear, actionable advice and warnings the UK relies on every day.
https://www.youtube.com/watch/-YHM-Sy4ihM?start=0