Atmosphere Ocean Science Colloquium

---

Insights into Forced Climate Change from Large Initial Condition Ensembles

Speaker: Paul Kushner, University of Toronto

Location: Warren Weaver Hall 1302

Date: Wednesday, February 25, 2015, 3:30 p.m.

Synopsis:

The use of large ensembles of multidecadal climate simulations that are identically forced but initiated from differing initial conditions is becoming increasingly widespread. We present two studies that rely on large ensembles to cleanly separate forced and internal extratropical climate. The use of large ensembles of multidecadal climate simulations that are identically forced but initiated from differing initial conditions is becoming increasingly widespread. We present two studies that rely on large ensembles to cleanly separate forced and internal extratropical climate variations in the recent past.The first study estimates the sea surface temperature (SST) response to anthropogenic forcing from observations over the last century, using a simple "pattern scaling" approach. The estimated SST spatial pattern of response includes warming of the mid-latitude coasts near the western boundary currents, the tropical Indian Ocean, and the Arctic and subarctic oceans. The large ensemble is used to test the robustness of the pattern; our estimate picks up about half of the structural variance of the anthropogenically forced SST pattern in observations. With additional simulations we use the anthropogenic SST pattern as a forcing that drives hydroclimate responses over land regions. These responses, particularly in North America and Africa, can be usefully compared to observed regional trends and to a limited extent can be separated from internal variability. The second study addresses the role of the Atlantic Meridional Overturning Circulation (AMOC) in driving predictable variations in North Atlantic sea surface temperature (NASST). Recent studies suggest that coupled climate models disagree on both the sign and the phasing of the correlation between AMOC and NASST indices. However, the AMOC-NASST relationship in these studies is contaminated by a forced temporally nonlinear signal that can be systematically removed using large ensembles. We argue that the apparent disagreement among models arises from a commingling of a "bottom-up" effect in which unforced AMOC variations lead to NASST variations of the same sign, and a "top-down" effect in which forced NASST changes lead to AMOC changes of the opposite sign. Once the forced variations are properly removed, the models exhibit much more consistent behaviour.