Quantifying Southern Annular Mode paleo-reconstruction skill in a model framework

Abstract

Past attempts to reconstruct the Southern Annular Mode (SAM) using paleo-archives have resulted in records which can differ significantly from one another prior to the window over which the proxies are calibrated. This study attempts to quantify not only the skill with which we may expect to reconstruct the SAM but also to assess the contribution of regional bias in proxy selection and the impact of non-stationary proxy–SAM teleconnections on a resulting reconstruction. This is achieved using a pseudoproxy framework with output from the GFDL CM2.1 global climate model. Reconstructions derived from precipitation fields perform better, with 89 % of the reconstructions calibrated over a 61 year window able to reproduce at least 50 % of the inter-annual variance in the SAM, as opposed to just 25 % for surface air temperature (SAT)-derived reconstructions. Non-stationarity of proxy–SAM teleconnections, as defined here, plays a small role in reconstructions, but the range in reconstruction skill is not negligible. Reconstructions are most likely to be skilful when proxies are sourced from a geographically broad region with a network size of at least 70 proxies.