Dr Philip Sansom
Associate Research Fellow
I am an applied statistician and my primary research interests are Bayesian modelling and uncertainty quantification in complex systems. Particular application areas include the climate system and infectious disease epidemiology.
In my PhD thesis, I developed methods for synthesising projections of future climate change from ensembles of multiple climate models and constrained by recent observations. Climate models do not output probabilities. Statistical methods are required in order to produce credible projections of future climate. Climate projections contain many sources of uncertainty. Policy makers require realistic assessments of these uncertainties in order to make informed decisions. Communicating this uncertainty to a wider audience is a challenging but important task in order to maintain public support for climate policy.
I am also interested in forecast verification and post-processing methods for weather and climate forecasts. Numerical weather prediction models suffer from a variety of biases and errors that can lead to unreliable forecasts. Statistical post-processing can improve the reliability of forecasts, benefiting both end-users and providers.
I am currently part of the EuroClim project, working in partnership with the Met Office to better understand the drivers of seasonal climate variability and improve seasonal forecast quality. I am currently developing methods for diagnosing sources of predictability in climate variables based on time-varying auto-regressive frameworks.
Prior to my PhD, I worked as a mathematical modeller for Public Health England. I am interested in various aspects of infectious disease epidemiology, including inference about key disease characteristics and transmission patterns, data assimilation and uncertainty quantification for computer simulations.