全球变化知识资源中心

Summer weather extremes are often associated with high-amplitude atmospheric planetary waves (Petoukhov et al., 2013). Such conditions lead to stationary weather patterns, triggering heat waves and sometimes prolonged intense rainfall. These wave events, referred to as periods of Quasi-Resonant Amplification (QRA), are relatively rare though and hence provide only a few data points in the meteorological record to analyse. Here, we use atmospheric models coupled to boundary conditions that have evolved slowly (i.e., climate), to supplement measurements. Specifically we assess altered probabilities of resonant episodes by employing a unique massive ensemble of atmosphere-only climate simulations to populate statistical distributions of event occurrence. We focus on amplified waves during the two most extreme European heat waves on record, in years 2003 and 2015 (Russo et al., 2015). These years are compared with other modelled recent years (1987-2011), and critically against a modelled world without climate change. We find that there are differences in the statistical characteristics of wave event likelihood between years, suggesting a strong dependence on the known and prescribed Sea Surface Temperature (SST) patterns. The differences are larger than those projected to have occurred under climate change since the pre-industrial period. However, this feature of small differences since pre-industrial is based on single large ensembles, with members consisting of a range of estimates of SST adjustment from pre-industrial to present. Such SST changes are from projections by a set of coupled atmosphere-ocean (AOGCM) climate models. When instead an ensemble for pre-industrial estimates is subdivided into simulations according to which AOGCM the SST changes are based on, we find differences in QRA occurrence. These differences suggest that to reliably estimate changes to extremes associated with altered amplification of planetary waves, and under future raised greenhouse gas concentrations, likely requires reductions in any spread of future modelled SST patterns.