Variations of Antarctic Oscillation during the past millennium and the twenty first century

Abstract:

The Antarctic Oscillation (AAO) profoundly affects the frequency and intensity of the synoptic weather events and precipitation/temperature in the mid- and high-latitudes Sothern Hemisphere. This study investigated the simulated changes in AAO during 850AD to 2100AD using fully coupled climate models participating in the fifth Coupled Model Intercomparison Project (CMIP5). The output from 11 models with specific historical anthropocentric and natural forcing during 850-1850AD (last millennium run) and 43 models during 1850-2005 (historical runs), as well as 40 models with 21^st century changes in greenhouse gases and anthropogenic aerosols following the Representative Concentration Pathway 4.5 (RCP4.5) and 8.5 (RCP8.5) for 2006-2100 are analyzed. Those models were classified into ‘good’ and ‘poor’ models based on their performance in simulating the observed AAO spatial pattern, interannual variability and trend. The AAO during 850-1850AD show considerable interannual variations. However, there are no noticeable decadal and centennial variations during this periods regardless of the models are good or poor. These results are inconsistent with the proxy-based AAO reconstructions and suggested that the models may underestimate the natural forcing during the Medieval Warm Period and the Little Ice Age. The multiple-model ensemble (MME) shows similar AAO variations as the reanalysis data during 1850-2005. The seasonal AAO has being steadily increasing before 1960s followed by abrupt increasing during 1970-2005. The increasing of AAO is projected to continue during the 21^st century. Under RCP4.5 scenario, the AAO will be weakly but steadily increasing until 2050s and then leveled off or slightly decreasing in the second half of this century. Under RCP8.5 scenario, the AAO will continue to increase in the 21^st century. Though the MME of the poor and good models show similar seasonal AAO variations during 1850-2005, the poor models projected much stronger AAO in the future under both scenarios compared to good models.