Heat waves, thunderstorms or floods are examples of extreme events. Investigating these events statistically using mathematical methods is both exciting and challenging. On the one hand, it is important to be able to predict the strength and frequency of such events as well as possible - after all, extreme events often have some high, typically negative, impact, are dangerous for people, cause damage to infrastructure or lead to major economic losses - on the other hand, this task is complicated by the fact that extreme events occur very rarely. In some cases it even happens that one wants to estimate probabilities of events that have not been observed so far: What is the risk of a city being flooded by a river despite protective measures? How often do you have to expect new record temperatures in summer? In order to answer such or similar questions, one has to "extrapolate" beyond the range of previously observed values. The field of extreme value theory or statistics provides the basis for doing so in a mathematically sound way.
What types of extreme events are you particularly interested in?
I am particularly interested in the statistical analysis of events that do not only occur
isolatedly at a specific point in time at a single location, but also have some temporal and
This applies especially to extreme events in the environment - whether storms, heavy rainfall, floods, heat waves or droughts. For all these events, one can ask how they proceed over time, how long they last or how large the affected area is. To answer this question, spatio-temporal dependencies must be considered. For this purpose, models of extreme value statistics can be applied in many ways.
What are the particular challenges?
Nowadays, large amounts of data are collected in many areas of application. A variety of relevant quantities and parameters are measured, often with high spatio-temporal resolutions. Thus, for a single extreme event not only a single measured number but a high-dimensional space-time field of measurement data is available. In order to analyze these data efficiently, suitable computationally intensive procedures and models are required. The models are too complex to solve many questions about the statistical properties of extreme events by analytical calculations. Instead, simulations that are as accurate and efficient as possible are required. These are topics that will engage my research group in the coming years. There will certainly be new and interesting collaborations. In addition to the cooperation within the department, interdisciplinarity plays a major role - after all, our work should also take into account the special requirements of individual applications. Especially the connection to the Cluster of Excellence SimTech offers
Many thanks for the interview.
Jun.-Prof. Marco Oesting
Head of the Research Group for Computational Statistics
Stuttgart Center for Simulation Science (SC SimTech)
& Institute for Stochastics and Applications