We are interested in fast, small scale atmospheric processes such as clouds, convection, and turbulence. We must improve our understanding of the two-way interaction between these fast processes and slower components of the climate system, including the ocean, land surface, and slower modes of atmospheric variability, to improve our estimates of future climate change. What is the role of fast phenomena in driving these slower processes? And how do these slower components in turn provide sources of predictability for the atmosphere? A particular focus of our group is on tropical climate, on timescales from days to centuries, studying phenomena such as deep convection, the El Nino-Southern Oscillation, and the Madden Julian Oscillation.
Weather forecasts and climate predictions are made using numerical models. Parametrisation schemes encode our physical understanding of small-scale processes in these models. However, the assumptions and simplifications made in parametrisation schemes are a large source of error and uncertainty in weather and climate predictions. Weather and climate predictions must therefore be made in a probabilistic manner. Our research focuses on characterising and reducing uncertainty in our predictions due to the representation of small-scale processes.
We combine a range of approaches to achieve these goals, including:
- global state-of-the-art weather and climate models;
- ultra-high-resolution limited area simulations that resolve processes of interest;
- observational datasets;
- theoretical advances and simple dynamical systems;
- machine learning