Integrating proxy and model estimates of Antarctic climate variability over the last millennium

Our understanding of how changes in global mean climate state will affect the Antarctic Ice Sheets is limited by the short duration of the instrumental period. Observational records routinely used to develop and evaluate predictive models are not long enough characterize the highly variable southern high latitudes. Here, we investigate the long-term drivers of Antarctic climate and sea ice dynamics by comparing paleoenvironmental reconstructions and simulations of late Holocene conditions (850 - 2005 CE) from the Community Earth System Model Last Millennium Ensemble (CESM-LME).

Last millennium cooling observed in proxy records from the Antarctic has a robust physical basis: in LME simulations global temperature decline between 850 CE and 1850 CE due to changes volcanic, solar, and orbital forcing. The contrast between a warmer "Medieval Climate Anomaly" (850 - 1250 CE) and cooler "Little Ice Age" (1450 - 1850) is most pronounced at the poles with an average of 0.55˚ C of cooling in the Arctic and 0.2˚ of cooling in the Antarctic in full forcing simulations. Differing feedbacks in the Arctic and Southern Oceans following major volcanic eruptions results in greater simulated cooling at the northern high latitudes.

However, the LME does not resolve the spatial pattern of warming and cooling observed in Antarctic proxy records. The LME simulates a continent-wide cooling trend over the last millennium while ice core reconstructions indicate that parts of Western Antarctica and the Eastern Ross Sea warmed since 1000 CE. We propose that both proxy uncertainties and model bias associated with the representation of the Southern Annular Mode likely contribute to the observed proxy-model mismatch.