RISK, RARE EVENTS AND EXTREMES / 1 July - 31 December 2009

The photograph, by Hervé Wadier, shows a powder-snow avalanche impacting the road to Roux (Abriès municipality, Hautes-Alpes, France) on 19th January 2004. The avalanche occurred after heavy snowfall and was triggered accidentally. 

Risks and extremes

Environmental change is expected to have a substantial impact on the timing and sizes of rare but potentially catastrophic events, such as extreme precipitation, avalanches, windstorms, high tides, and heat waves. Although their consequences can be enormous, such events are very difficult to predict based on past data, and statistical thinking is essential. Yet current statistical and mathematical tools for modelling extreme events are inadequate to deal with this task. There is thus an urgent need to improve current models for statistical extremes, and to develop new methodology, in order to better understand, predict, and manage these risks.
 
The best-developed and most important mathematical models for rare events are based on probabilistic models for extremes, fitted to data using statistical techniques. The available data are often scarce, because such events are necessarily unusual, and careful and sophisticated modelling is needed to extract the fullest information from the data, and to provide realistic forecasts and associated measures of uncertainty.

Extremal modelling for single stationary time series is now well-established and is increasingly widely used in environmental and engineering problems, but methods for multiple time series and spatial data are less well-developed, and a major effort is needed to develop approaches for dealing with realistic problems, in applications where spatio-temporal variation and nonstationarity are key elements.

The scientific goal of this programme is to push this effort forward by bringing together major researchers in statistics of extremes and specialists from domains to which advances in extremal modelling could contribute in an essential way. Although the focus of the programme will be on environmental extremes, there is considerable potential for cross-pollination with domains such as  finance and telecommunication systems, where related models are used.

Programme activities

-  a kick-off workshop on Spatial Extremes and Applications, from 13-17 July 2009, which brought together researchers for whom modelling of spatial extremes is crucial, such as in climatology and insurance.  Participants included: Juliette Blanchet, Adri Buishand, Dan Cooley, Laurens de Haan, Emma Eastoe, Ana Ferreira, Petra Friederichs, Sophie Fukutome, Ivette Gomes, Armelle Guillou, Jurg Hüsler, Janet Heffernan, Richard Katz, Christian Lantuejoul, Robert Lund, Douglas Maraun, Simone Padoan, Mathieu Ribatet, Holger Rootzén, Martin Schlather, Richard Smith, Feridun Turkman, Cristiano Varin, Max Werner; 

-  a workshop on High-dimensional Extremes, from 14-18 September 2009, including applications to environmental science, finance and other areas.  Provisional participants include: August Balkema, John Einmahl, Paul Embrechts,  Ana Ferreira, Anne-Laure Fougères, Sidney Resnick, David Walshaw, Zhengjun Zhang;

-  a workshop on Spatio-temporal Extremes and Applications, from 9-11 November 2009, with a view to modelling of dynamic aspects of extremes, for example in hydrology, rainfall analysis, flood studies, etc.  Provisional participants include: Clive Anderson, Paola Bortot, Richard Davis, Chris Ferro, Arnoldo Frigessi, Carlo Gaetan, Alan Gelfand, Rick Katz, Huiyan Sang, Martin Schlather, Feridun Turkman; and 
-  a Final Conference, from 12-13 November 2009, bringing together the statistical and various user communities, to present the new state of the art in both theory and applications of extremal modelling; and 

-  an active visitor and seminar programme under which junior and senior researchers will work at EPFL for varying periods.

Scientific organising committee