Mitchell Lecture - Evolving extremes: Statistical models for spatial and non-stationary environmental risk assessment
How can we assess the risk of floods and extreme rainfall beyond anything recorded in the past? Professor Thordis L. Thorarinsdottir explores how statistics and machine learning help us understand evolving environmental hazards in a changing climate. Discover how researchers model rare events across different places and times—and the challenges that remain. Join us on Monday 5 October 2026, 4pm with a wine reception from 5pm
Distinguished Lecture Series in Mathematics and Statistics
Date: Monday 05 October 2026
Time: 16:00 - 17:00
Venue: 311B Maths & Stats
Category: Public lectures, Social events, Academic events, Student events, Alumni events, Staff workshops and seminars, Community events
Speaker: Thordis L. Thorarinsdottir
Mitchell Lecture 2026
Evolving extremes
Statistical models for spatial and non-stationary environmental risk assessment
Professor Thordis L. Thorarinsdottir
University of Oslo
Attendance is free, but advance registration is required:
https://mitchell-lecture-2026.eventbrite.co.uk
How can we assess the risk of floods and extreme rainfall when the most severe future events may have no precedent in the historical record?
The 2026 Mitchell Lecture will be delivered by Professor Thordis L. Thorarinsdottir of the University of Oslo. She will explore how modern statistical methods can help us understand and prepare for rare environmental events in a changing climate.
Assessing environmental risk requires researchers to estimate the likelihood and severity of events more extreme than those previously observed. Climate change makes this task still more difficult: the physical processes generating rainfall and flooding are changing over time, while environmental hazards are connected across different locations and timescales.
Professor Thorarinsdottir will introduce statistical frameworks designed to address these challenges. The lecture will consider how concurrent risks can be assessed across multiple locations and how hazards can be modelled over durations ranging from short, intense rainfall to prolonged periods of flooding. It will also examine the methodological and computational compromises involved in developing useful models, drawing on applications to real-world hydro-meteorological data.
The lecture will conclude by considering the limitations of current approaches and the open questions requiring new statistical ideas.
Professor Thorarinsdottir is Professor of Statistics and Data Science and Deputy Head of the Department of Mathematics at the University of Oslo. Her research spans stochastic modelling, machine learning, uncertainty quantification and probabilistic prediction, with a particular focus on applications in environmental science. She is also a Joint Editor of the Journal of the Royal Statistical Society: Series C (Applied Statistics).
The lecture will take place at 4pm on Monday 5 October 2026 in Lecture Room 311B of the Mathematics and Statistics Building at the University of Glasgow. A wine reception will follow at 5pm.