The large-scale atmospheric circulation response to climate change drivers: a multi-model comparison study

In mid-latitudes, changes in large-scale atmospheric circulation can substantially alter regional climate including the likelihood of extreme events. Over recent decades, robust changes in circulation have been identified in observations, including a poleward shift of the summertime Southern Hemisphere (SH) jet and widening of the tropical belt. The midlatitude circulation is sensitive to external forcing and climate models project significant future trends under high radiative forcing scenarios. However, there remain significant uncertainties in these responses, including the relative importance of different external drivers, the physical mechanisms that underpin responses and the robustness of responses in ensembles of climate models. This thesis aims to advance understanding through multi-model assessments of large-scale tropospheric circulation responses to CO2 and non-CO2 drivers by investigating dynamical sensitivity, rapid adjustments and sea-surface temperature (SST) mediated responses, and sources of uncertainty arising from structural model differences and internal variability, focusing on the SH. SST-mediated feedbacks are shown to be the most important component of the large-scale response to greenhouse gases, scattering aerosols and solar forcing, with the evolution of SST warming patterns in the initial decade after abrupt forcing perturbations of particular importance. Rapid adjustments also contribute a substantial fraction of the midlatitude circulation response, especially to black carbon (BC) aerosols, with 75% of the SH jet shift to a 10xBC perturbation mediated by direct tropospheric heating anomalies. 80-90% of the intermodel difference in summertime midlatitude low-level zonal wind responses can be explained by differences in tropics-to-pole temperature gradient responses. A role for anthropogenic aerosols in recent tropical width trends is suggested. However, ensemble sizes in current generations of Coupled Model Intercomparison Project (CMIP) models are too small to robustly detect and quantify this response. The results motivate development of large ensemble single forcing simulations to improve modelled representation of historical and projected future circulation shifts.