Upgrading climate predictions with AI

19th December 2018


Wadham Fellow Laure Zanna aims to improve the accuracy of climate change predictions using AI.

  • Image showing ocean turbulenceTurbulence in an ocean climate model (from Kjellsson and Zanna, 2017)

    Turbulence in an ocean climate model (from Kjellsson and Zanna, 2017) 

“Our work will improve the way climate change predictions are made,” says Wadham’s David Richards Fellow, Laure Zanna. “Policy makers can use these more accurate predictions to create more effective climate change policy,” she adds.

The Oxford Climate and Ocean Physics team, which Laure leads, aims to perfect the numerical models used to explain changes to both ocean and climate systems. Much of their recent work is done in collaboration with environmental modelling centres, whose predictions define global policy decisions on climate issues. 

Laure’s work focuses on understanding the role of oceans in climate change. As carbon is emitted into the atmosphere, sea temperatures rise, with much of the extra heat being absorbed by the oceans. This leads to increases in sea levels and changes to water currents. Laure’s team uses equations, data and simulations to create models explaining ocean and climate changes historically, in order to predict future changes. 

However, although data are plentiful (sea temperature records go back to 1870, for example) there are still gaps. Subsequently, one new way that Laure’s team improves current ocean modelling methods is by using data from observations or numerical models together with Artificial Intelligence (AI). “The data are imperfect,” says Laure, “We can improve the accuracy of predictions by feeding the data (from observations or numerical simulations) into machine learning algorithms to predict physical processes, which cannot be directly observed or resolved in numerical simulations.” 

Another development for the team is their prediction that there will be regional variations in levels of ocean warming, at least up until 2050, including the cooling of some areas. “The ocean takes up heat and redistributes it,” says Laure. “Some areas might get warmer while other can get colder affecting for example the survival of coral reefs. This area of research opened up a lot of new questions for our team.” 

Laure was born in France and grew up in Marseilles before spending time in Israel. She then took her PhD in Climate Dynamics at Harvard University before joining Oxford initially as an Oxford Martin Fellow and then an Associate Professor in the Department of Physics. She was a Junior Research Fellow at Balliol College from 2009 to 2011, a lecturer at Christ Church from 2014 to 2017, and a Fellow of St Cross from 2011 to 2018. 

And with nine years of Oxford academic life behind her, Laure is reaching an exciting stage in her research. “Now I really want to push our ideas even further,” she says. “I want to bring machine learning and a proper theory of physics for climate change together in a way that will improve the climate projections and models of the future.”