I work on modelling atmospheric time series data using stochastic differential equations.
My research is focused upon developing reduced complexity models of atmospheric dynamics. I apply stochastic modelling techniques and Bayesian inference methods to model atmospheric low-frequency variability. This has potential benefits for long-term weather predictability and improved coupling of atmospheric processes to ocean simulation models.
More specifically I am studying a stochastic mode reduction strategy1,2 that reduces a high dimensional model to a stochastic model for the variables of interest alone. This method is based upon averaging out the fast modes of variability to produce a model for the low frequency modes. One has to parametrise the unresolved variables with a tractable form and then infer the unknown parameters from data. For this I am using recent advances in Bayesian inference methodology for stochastic differential equations3.
I am currently focussed upon techniques to include prior information on unknown parameters in order to ensure that the resulting dynamical system has stable (non-transient) solutions.
Peavoy, D.; Franzke, C.. 2010 Bayesian analysis of rapid climate change during the last glacial using Greenland δ18O data. Climate of the Past, 6(6). 787-794
Previously I have worked upon statistical modelling of Greenland ice core data to determine whether there existed any periodicity in the rapid climate change events that occured during the last Glacial preiod.
I did my undergraduate degree in theoretical physics at Liverpool University. Here is my undergraduate project
I've also completed an MSc in Natural Computation in the computer science department at Birmingham University, where I studied Bayesian Inference of genetic networks. Here is a poster outlining this research