Weather Prediction, Numerical or otherwise

Colm Clancy (University College Dublin)

Time:

2PM Wednesday, 18 February 2026

Location:

Science East, E0.32

The development of Numerical Weather Prediction has been referred to as a "quiet revolution", as advances in the science and computing over the past few decades have led a slow, but consistent, improvement in the accuracy of weather forecasts. As operational models now target resolutions on the order of a kilometre or below, new challenges arise, not least in terms of model efficiency and numerical stability. On the other hand, there has recently been a rapid development and deployment of data-driven models which claim to outperform traditional physics-based models. In this talk I will give an overview of my recent work in sub-kilometre modelling, as well as discussing the future research challenges and opportunities facing the field.