Attribution of, and constraints on, changes in extreme precipitation under climate change

This thesis investigates changes in the characteristics of extreme precipitation, under a regime of climate change. This understanding is important, as it is often stated that the impacts of climate change will be felt most through changes in extremes, and there are serious impacts projected for the 21st century. The Intergovernmental Panel on Climate Change also recently highlighted that understanding the statistical distribution of precipitation is key to understanding future changes in its extremes. Concurrently, however, it identified weaknesses in studies of extreme precipitation which were due to availability, quality and definition of data.

The first half of this thesis investigates a concept that could potentially simplify this problem. We develop work of a previous study which argued that, under global warming, the Clausius-Clapeyron relation for atmospheric moisture availability provides a physical quantifiable constraint on changes in extreme precipitation. Moreover, it was argued this relation may provide a tighter constraint on such changes than would be inferred directly from observations or models, since it is dependent solely on temperature which is historically the best represented climate variable. By searching, in a climate model under global warming, for the emergence of this thermodynamic constraint over varying regional and seasonal scales we find evidence for its emergence only in mid-latitude regions, with its prediction of extreme precipitation change being better than if the change in mean precipitation had been used directly from the model. The conclusion is that in regions where there is no emergence, changes in the nature of the atmospheric dynamics must also be considered.

Motivated by this need for a fuller understanding, the second half of this thesis turns to the potentially more valuable question of attributing the risk of specific extreme precipitation events to anthropogenic climate change -- with a view to societal mitigation of and/or adaptation to such events. We focus on the United Kingdom floods of Autumn 2000, which occurred during the wettest autumn on record, and describe the concept of Probabilistic Event Attribution that enables us to asses the extent to which the risk of such an event is attributable to 20th century emissions of anthropogenic greenhouse gases. Using large ensembles of simulations generated with a high spatio-temporal resolution climate model, the comparison is made between an Industrial Autumn 2000 climate which includes the effects of atmospheric greenhouse gases for that time, and a Non-industrial climate had there been no emissions of these gases over the 20th century. We find that the Industrial climate is able to capture Autumn 2000 like precipitation and that there is some evidence for attribution of such an occurrence to the aforementioned greenhouse gases. The latter conclusion, however, is dependent of on the nature of the sea surface temperatures input to the Non-industrial climate model being compared against, and this requires further investigation.

Related information is available at http://attribution.cpdn.org.
Email address: p.pall@atm.ox.ac.uk