Sophia Wisniewska

Profile

I am PhD student in the Institute for Climate and Atmospheric Science (ICAS). I am funded by the NERC UNRISK CDT and hold a CASE partnership with the Met Office.

Before starting my PhD, I graduated in a Physics (MSci) from the University of Bristol, and worked for four years as a data science consultant in London. Through my data science role, I gained hands-on experience working with diverse datasets, using machine learning and data science techniques to gain insight from company data to drive business decisions. In my spare time, I enjoyed reading about atmospheric science, and gained a particular interest in cloud processes. This motivated me to conduct research in climate modelling to reduce the uncertainty in climate predictions – at a time when fostering global commitment to reducing the impact of anthropogenic climate change is more critical than ever.

Research interests

Cloud behaviour is governed by a wide range of complex processes occurring across multiple scales, from microphysical interactions within clouds to large-scale cloud dynamics. Cloud feedbacks and cloud-aerosol interactions are still poorly represented in climate models, and are the largest source of uncertainty in climate sensitivity estimates. Many cloud processes occur on scales smaller than the model grid resolution, and must be approximated through parametrisations. As seen in coupled model intercomparison projects, the intermodel spread in equlibrium climate sensitivity is due to differences in how the models represent clouds through parameterisation schemes. There is also uncertainty – so-called ‘parametric uncertainty’ that is associated with the uncertain parameters themselves.

My PhD research at the University of Leeds employs a novel methodology using Perturbed Parameter Ensembles (PPEs) to train machine learning emulators. This approach enables efficient exploration of parameter space and to understand the joint effects of the parameters that drive cloud feedback and cloud adjustments, and quantify the parametric uncertainty. Observational data may then be used to constrain the uncertainty, with the goal of improving the representation of clouds in climate models, and ultimately reduce uncertainty in climate sensitivity estimates.

Qualifications

• MSci Physics, University of Bristol (2020)