This provides significant efficiency and performance improvements for most sectors. These include how to predict the weather and its impact.
As a national weather service in the UK, Met Office has taken steps to accelerate the adoption of these technologies and integration through climate science and services. Our Development Ability, people, partnerships [LINK] Through intensive engagement with organizations that apply a variety of pilot projects, training activities, and data science to support the ongoing delivery of extraordinary benefits and values.
Why is AI prediction necessary?
Over the past decades, advances in technology have increased the power and refinement of computer models that support weather and climate science and services. Recent advances in these areas have led the world to the pinnacle of breakthroughs that could change the game in weather and climate modeling. AI and ML have the potential to promote new advances in weather and climate science to ensure society can be safe and prosperous.
This is an important, timely opportunity that could help us tackle one of the most serious threats facing today's civilization. Our vulnerability to extreme weather phenomena in changing climates – science and technology that allows us to better understand and manage important dangers play a key role in increasing climate resilience through both mitigation and adaptation.
How can AI enhance forecasting?
Met Office has always been a data science organization and has been active in some degree of AI and ML, but new techniques in recent years (particularly deep learning) have shown impressive success in many domains, such as computer vision and natural language processing, indicating the wider application potential.
This is true throughout the weather and climate science and services process, through observations (including quality control and gap filling), simulations (including data assimilation and model simulations), and analysis (including post-processing), products and services (including risk forecasting and warnings).
Advances in AI and ML provide the possibility to not only significantly enhance existing physics-based weather models by incorporating the elements of AI and ML, but also complement the use of AI-based weather models that function in a completely different way than physics-based models.
Physics-based and AI-based models
Physics-based weather forecasting models such as operations Momentum modelwhich runs on a large supercomputer and relies on solving complex equations (representing the management physics of fluid dynamics) to calculate how atmospheric conditions change over time from what is currently being observed.
In contrast, AI weather prediction models such as experiments FastNet Model (Developed by Met Office and The Alan Turing Institute) directly solves the management physics of fluid dynamics, and instead provides an alternative to learning patterns from a very large amount of historical data (such as observations and physics-based simulations). By leveraging a vast amount of information, AI models attempt to identify and utilize unique patterns of data that are not explicitly expressed.
Once trained, these AI-based models offer great advantages in the computational power required at runtime and the potential to improve prediction performance and resolution.
However, potential drawbacks include a lack of understanding of physics. This can make it difficult to predict extreme weather events that become more frequent as a result of climate change. So next-generation predictions could include a mix of AI-based and physics-based prediction methods, and meteorologists continue to play a key role in determining the most ridiculous weather in the coming days and weeks.
Similarly, because of this, supercomputers not only continue to run physical-based predictions themselves, they remain as important as ever when creating parts of the training dataset on which AI-based predictions rely.
The Continuing Role of Meteorologists
Meteorologists will continue to play a key role in verifying and publishing predictions and guidance.
The advent of AI-based forecasts could revolutionize operational weather in much the same way as computers when Met Office began to produce numerical weather forecasts in 1965.
Since then, as is now, operational meteorologists and subject matter experts continue to be an integral and important component of the forecast and warning process, playing a key role in analyzing output, understanding uncertainty, and communicating trustworthy weather insights and intelligence to customers and communities.
Met Office employs the use of AI across the workforce to ensure an efficient, effective and future organization as a UK national weather service. You can find out more about it I met AI and the people in the office here.
