Importance of Framing for Extreme Event Attribution: The Role of Spatial and Temporal Scales

Event attribution, which determines how anthropogenic climate change has affected the likelihood of certain types of extreme events, is of broad interest to industries, governments, and the public. Attribution results can be highly dependent on the definition of the event and the characteristics assessed, which are part of framing the attribution question. Despite a widely acknowledged sensitivity to framing, little work has been done to document the impacts on attribution and the resulting implications. Here, we use a perfect-model approach and large ensembles of coupled climate-model simulations to demonstrate how event attribution depends on the spatial and temporal scales used to define the event. In general, stronger attribution is found for events defined over longer time scales and larger spatial scales due to enhanced signal-to-noise ratios. With strong warming trends, most regions see large changes in the likelihood of temperature extremes at all scales, even at low levels of global mean temperature increase. For precipitation extremes, spatial scale plays a strong role. It may be possible to attribute changes in likelihood for extreme precipitation events defined over larger scales, but greater levels of global warming are often required before it is possible to attribute changes in the likelihood of smaller-scale precipitation events. Care must be taken to understand the scales used in event attribution, in order to properly understand the results.