The UK Met Office data portal publishes 500x500 pixel radar images of precipitation intensity across the UK, updated in 15-minute intervals. The spatial distribution of rainfall patterns, and their evolution in time, can be used to skilfully predict rainfall a few hours ahead. This data-driven approach to short-term forecasting is called radar nowcasting. This project will develop and implement statistcal learning techniques for radar nowcasting across the UK. The student will gain expertise in modern statistical modelling methods for large spatio-temporal data sets, as well as the computational tools to implement these methods efficiently. A potential application of this project is a real-time web service that issues routine short-term rain predictions at locations across the UK.