Attribution of UK Winter Floods to Anthropogenic Forcing

Thursday, 18 December 2014
Nathalie Schaller1, Kay Alison2, Sarah Naomi Sparrow1, Friederike Elly Luise Otto1, Neil Massey1, Robert Vautard3, Pascal Yiou3, Geert Jan van Oldenborgh4, Ronald van Haren4, Rob Lamb5, Chris Huntingford2, Sue Crooks2, Tim Legg6, Antje Weisheimer1, Andy Bowery1, Jonathan Miller1, Richard Jones6, Peter Stott7 and Myles Robert Allen8, (1)University of Oxford, Oxford, United Kingdom, (2)Centre for Ecology and Hydrology, Wallingford, United Kingdom, (3)LSCE Laboratoire des Sciences du Climat et de l'Environnement, Gif-Sur-Yvette Cedex, France, (4)Royal Netherlands Meteorological Institute, De Bilt, Netherlands, (5)JBA Trust, Skipton, United Kingdom, (6)Met Office Hadley center for Climate Change, Exeter, United Kingdom, (7)Met Office Hadley center for Climate Change, Exeter, EX1, United Kingdom, (8)University of Oxford, ECI/School of Geography and the Environment, Oxford, United Kingdom
Many regions of southern UK experienced severe flooding during the 2013/2014 winter. Simultaneously, large areas in the USA and Canada were struck by prolonged cold weather. At the time, the media and public asked whether the general rainy conditions over northern Europe and the cold weather over North America were caused by climate change. Providing an answer to this question is not trivial, but recent studies show that probabilistic event attribution is feasible.
Using the citizen science project weather@home, we ran over 40’000 perturbed initial condition simulations of the 2013/2014 winter. These simulations fall into two categories: one set aims at simulating the world with climate change using observed sea surface temperatures while the second set is run with sea surface temperatures corresponding to a world that might have been without climate change. The relevant modelled variables are then downscaled by a hydrological model to obtain river flows. First results show that anthropogenic climate change led to a small but significant increase in the fractional attributable risk for 30-days peak flows for the river Thames.
A single number can summarize the final result from probabilistic attribution studies indicating, for example, an increase, decrease or no change to the risk of the event occurring. However, communicating this to the public, media and other scientists remains challenging. The assumptions made in the chain of models used need to be explained. In addition, extreme events, like the UK floods of the 2013/2014 winter, are usually caused by a range of factors. While heavy precipitation events can be caused by dynamic and/or thermodynamic processes, floods occur only partly as a response to heavy precipitation. Depending on the catchment, they can be largely due to soil properties and conditions of the previous months. Probabilistic attribution studies are multidisciplinary and therefore all aspects need to be communicated properly.