The future of climate modelling

Observations over past decades, and current monitoring of the climate, show unequivocally that the Earth’s climate is changing and that the impacts of that change are being felt in myriad ways – from more intense heatwaves, accelerating sea level rise, increasing rainfall intensity, wildfires, and ice loss. The principal lens that scientists use to understand these changes, and to make predictions of future change, are climate models. These numerical simulations of the physics, chemistry and (increasingly) the biology of the climate system have a 40-year track record of successful predictions. However, while their complexity has grown and their skill has improved over the decades, they remain incomplete and insufficiently detailed. Meanwhile, the questions that are being posed by decision makers and the general public are challenging both the theoretical basis of the models, their capacity to deliver answers, and the infrastructure that the scientific community has built around them. To be frank, we are being asked questions about climate change in real time that we cannot answer. I will discuss where we find ourselves in 2024 where climate records are being broken on a sometimes daily basis, but where the opportunities to rethink what we are doing have never been greater. Does the future lie in ultra-high resolution, or machine learning, or greater complexity, or more operational stances, or a combination of all of these approaches? How will this be managed and who will make these decisions? I hope this lecture will provide some necessary background for this continuing conversation.