The University of Exeter and IBM are partnering on a new PHD Studentship, which will explore Bayesian methods for Machine Learning and Artificial Intelligence to help accurately predict the impacts of climate change.
The new project will connect the University of Exeter’s experts in environmental intelligence with IBM Research, whose Climate and Sustainability research programme is currently exploring how AI can be used to help build resilience to climate change by quantifying the risk of extreme climate events.
One area of focus is the development of data-driven climate impact models which use traditional AI methods and a rapidly increasing volume of in-situ and remote-sensed earth observation data, to predict the impacts of climate change.
The new studentship aims to overcome a key gap in this research to date – namely, the need to develop a methodology which better accounts for the variety across climate impact models and data, especially with regard to ‘uncertainty’. Failing to correctly account for uncertainties when building and deploying climate models can result in over-confident predictions, which are at risk of being considered unreliable and inaccurate.
The project will address this gap using Bayesian hierarchical modelling, which has the potential to enable hybrid climate models to produce more reliable and well-calibrated probabilities of high-impact climate events.
The new methodology will then be tested for real world applications in development at IBM, improving prediction of climate hazards such as floods and heatwaves, and their impact on buildings, energy networks, transport and communications.
“Accurate, locally-precise predictions of the impacts of climate change are urgently needed across multiple commercial sectors”, says Dr Anne Jones, Senior Research Scientist in Climate and Sustainability at IBM Research, “such as supply chain, financial services, energy, utilities, and infrastructure management.”
“But currently, many of the sources of uncertainty involved in such predictions are completely ignored. Using the methodology developed in the project,” she continues, “we will able to provide organisations with better information about where to focus their resources, enabling much more efficient climate resilience planning.”
The studentship will be delivered through the University of Exeter’s Centre for Doctoral Training (CDT) in Environmental Intelligence. The Centre represents a step change in environmental education and research, by taking a holistic, interdisciplinary approach when training environmental scientists.
Professor Hywel Williams, Associate Professor in Data Science at the University of Exeter and Director of the Centre, says “the studentship is an exciting example of the value of interdisciplinary approaches to climate change, which harness our expertise in environmental intelligence to make real solutions possible.”
“We are excited to partner with IBM on this project, which shows how the Centre for Doctoral Training can deliver environmental intelligence training needed to meet the challenge of adapting and building resilience to climate risks.”