physics | climate | python


Forecast-based attribution of extreme weather events

The primary focus of my research looks into how climate change is affecting individual extreme weather events. Due to the fact that no directly observable world without human influence on the climate exists, all approaches attempting to answer this question involve some form of modelling, either numerical or statistical. In general, extreme event attribution studies now tend to rely on relatively coarse climate models, and use several different approaches or models to assess the robustness of their conclusions. My work aims to adapt weather forecast models that are demonstrably able to simulate the actual weather extreme being studied to allow them to answer this same question. You can read more about why I am doing this here.


A collaboration with the Met Office and climateprediction.net, the aim of this project is to understand the uncertainty surrounding future extreme winter weather in order to support climate projection efforts such as the UK Climate Projections (UKCP18). I have been heavily involved in analysis of the model output coming out of this project and in writing the accompanying science paper. You can read more about this project here, and the science paper is available here.


Through my masters project exploring the remaining carbon budget under ambitious mitigation scenarios, I became familiar with reduced complexity climate models / emulators, and in particular, the FaIR climate model. Later on, during the early stages of my PhD, I worked on updating FaIR to make it quicker, simpler and more transparent - attempting to identify a minimal level of structural complexity possible. The result was FaIRv2.0.0, a set of six equations that can be used to model the 1-D globally averaged response of the climate system to greenhouse gas and aerosol emissions. Its speed and flexibility lends it to be used for large ensembles of perturbed parameter probabilistic projections or for emulating the response of more complex models (such as those contributing to CMIP6). You can read more about this model update here.