Dr Jill Johnson

Position: Lecturer
Office: I11 Hicks building
Photo of Jill Johnson


MAS113 Introduction to Probability and Statistics Information  


Interests: Uncertainty quantification for computer models; Surrogate modelling (emulation); Model-observation comparison (history matching).
Research group: Statistics


Assessment administrator


Dr Johnson studied at Newcastle University, graduating in 2010 with a PhD in Statistics. Her thesis was titled ‘Modelling Dependence in Extreme Environmental Events’. Following this, she worked in the civil service as a research statistician at the government’s Food and Environment Research Agency, looking at uncertainty quantification and risk analysis for multiple applications including food safety and land use change.

In December 2012, Dr Johnson returned to academia as a research associate in the aerosol research group at the Institute for Climate and Atmospheric Science, University of Leeds, where her work focussed on the quantification and constraint of key uncertainties in complex models of the atmosphere and climate.

In August 2021 she joined the School of Mathematics and Statistics as a Lecturer in Statistics.

Research interests:

Dr Johnson’s research interests are in the development and practical application of statistical methods to quantify, assess and then reduce uncertainty in large-scale complex models of real-world systems. To date, she has focussed on problems in environmental science and particularly in relation to the atmosphere and climate: She has worked with a range of models on different scales including the simulation of an individual cloud, to the simulation of a cloud field, to the simulation of the global distribution of aerosols in the atmosphere. Her approaches include expert elicitation to inform parameter choices, the use of surrogate statistical models (Gaussian process emulation) to enable dense sampling over a complex model’s input uncertainties, sensitivity analysis to understand the driving sources of uncertainty and ‘history matching’ to reduce uncertainty via comparison to observations. In her research, Dr Johnson aims to address the statistical challenges of model-observation comparison to constrain the effects of high-dimensional model parameter uncertainty when using diverse observations that have sparse spatial and temporal coverage, and to explore model uncertainty when a model output is non-stationary.