AI: change in weather forecasting | Insights and research

Climate change has led to a growing number of extreme weather events around the world. This year alone those events have included everything from a record-breaking heatwave across large parts of Europe to one of the most powerful hurricanes ever to make landfall in the US and catastrophic drought in Somalia.

Better weather forecasting could play an essential role in predicting such events so that countries can prepare. People could be evacuated from the worst affected regions, for example, or preparations could be made to supply water to areas where it will become scarce.

Weather forecasting is notoriously difficult but artificial intelligence and machine learning could hold the key

With advances in data science, computers can now analyse and learn from vast volumes of information at speed and with a high level of accuracy. To take advantage of these technological breakthroughs, weather and climate science organisations across the globe are revising their operating plans and require increasing numbers of data scientists to meet their needs.

In 2020, the National Oceanic and Atmospheric Administration (NOAA) in the US published its AI Strategy, dramatically expanding ‘the application of artificial intelligence in every NOAA mission area’. In 2021, the European Centre for Medium-Range Weather Forecasts (ECMWF) published a roadmap for the deployment of machine learning for weather and climate prediction, while a research theme on fusing simulation with data science is embedded within the UK Met Office’s 2022 Research and Innovation Strategy.

STEM skills

In a recent Met Office report Chief Science and Technology Officer Professor Stephen Belcher, said: “Over the last few decades, we have seen advances in technology increase the power and sophistication of computer models that underpin weather and climate science and services. Recent progress in these areas has brought the world to the cusp of potentially game-changing breakthroughs in weather and climate modelling.

“Data science and artificial intelligence have huge potential to drive forward new advances in weather and climate science to help make society better able to survive and thrive in a changing climate.”

Forecasting changes in the weather requires vast amounts of observational data, advanced understanding of complex, physics-based models and access to significant computational power. Fusing data science with conventional physical modelling and expert knowledge is expected to lead to improvements in computational efficiency, information quality and interpretation.

The Met Office report, however, highlights the challenge the organisation faces in recruiting the experts it needs, saying: “There is growing competition for STEM skills, in particular data science skills, which are fundamental to the success of the Met Office.” It goes on to describe its strategy of investing in its people and culture to attract and retain the best data scientists.

Revolutionising prediction

In Europe, the European Weather Cloud, which is being developed by ECMWF and the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT is expected to play a significant role in the development of machine learning tools. It will enable researchers to easily load data from the data archive and to use standard machine learning software that is very different from the tools typically used in ECMWF’s supercomputing environments.

A recent newsletter from the organisation said: “Many standard methods used by ECMWF scientists on a daily basis can be regarded as examples of machine learning. However, there has recently been a surge in new methods which have the potential to revolutionise the work of operational weather prediction centres.”

But while developed nations are investing, the technology required to predict weather more accurately is too expensive for many of the countries that are worst affected by climate change. That could be about to change, however, with tech companies competing to develop better systems at lower cost. Uganda’s National Meteorological Authority, for example, has acquired an AI system from a California-based firm that is said to be considerably cheaper than others on the market.

Climate resilience

Many other organisations are also becoming involved in efforts to accurately predict the weather. Researchers led by scientists at NASA’s Jet Propulsion Laboratory have used machine learning to develop an experimental computer model that can more accurately detect the intensification of storms into hurricane like Hurricane Ian, which struck Florida this year. Google’s London-based Deep Mind AI company has also been working to improve the accuracy of short-term weather forecasts.

As the Met Office report, which was published before the most extreme events of 2022, says:

Science and technology are the essential tools in the fight against climate change

It explains: “Extreme weather events, such as in 2021 — a record heatwave on the west coast of Canada, catastrophic floods in Belgium, Germany and the Netherlands, or wildfires in several countries of southern Europe — all highlight our vulnerability to natural climate variability in the present-day and are a forewarning of the increasing risks associated with climate change. The need for action on climate change is urgent; science and technology have a vital role to play in ensuring we can better understand and manage the key hazards in order to become climate resilient through both mitigation and adaption.”

STEM skills are becoming increasingly critical in weather forecasting, as extreme weather events become more frequent, and in the fight against climate change more broadly. The expertise of STEM professional is badly needed everywhere and national meteorological organisations across the globe are seeking out expertise in data science, AI and machine learning.