Prediction and Evaluation

Our research aims to improve weather and climate predictions by developing better ways of constructing probabilistic predictions and of evaluating the performance of prediction systems. A notable feature of our research is our focus on predictions of rare events, which often have the most severe impacts on the world.

Professor Peter Challenor Professor
Professor Mat Collins Professor
Dr Christopher Ferro Senior Lecturer
Dr Tim Jupp Lecturer
Professor David B. Stephenson Professor
Dr Daniel Williamson Lecturer and EPSRC fellow

Our prediction work focuses on generating and post-processing ensembles of projections from numerical weather and climate models. Our evaluation work includes methods for assessing predictions ranging from short-term weather forecasts to decadal climate predictions, for predicting how changing a prediction system will affect its performance, and for understanding the physical processes underlying predictive performance.

The methods that we develop rely on exploiting patterns in the relationships between model projections, predictions and observations, which means that statistical modelling underpins all of our research.