Antarctic temperature and CO₂: near-synchrony yet variable phasing during the last deglaciation

Abstract

The last deglaciation, which occurred from 18 000 to 11 000 years ago, is the most recent large natural climatic variation of global extent. With accurately dated paleoclimate records, we can investigate the timings of related variables in the climate system during this major transition. Here, we use an accurate relative chronology to compare temperature proxy data and global atmospheric CO2 as recorded in Antarctic ice cores. In addition to five regional records, we compare a δ18O stack, representing Antarctic climate variations with the high-resolution robustly dated WAIS Divide CO2 record (West Antarctic Ice Sheet). We assess the CO2 and Antarctic temperature phase relationship using a stochastic method to accurately identify the probable timings of changes in their trends. Four coherent changes are identified for the two series, and synchrony between CO2 and temperature is within the 95 % uncertainty range for all of the changes except the end of glacial termination 1 (T1). During the onset of the last deglaciation at 18 ka and the deglaciation end at 11.5 ka, Antarctic temperature most likely led CO2 by several centuries (by 570 years, within a range of 127 to 751 years, 68 % probability, at the T1 onset; and by 532 years, within a range of 337 to 629 years, 68 % probability, at the deglaciation end). At 14.4 ka, the onset of the Antarctic Cold Reversal (ACR) period, our results do not show a clear lead or lag (Antarctic temperature leads by 50 years, within a range of −137 to 376 years, 68 % probability). The same is true at the end of the ACR (CO2 leads by 65 years, within a range of 211 to 117 years, 68 % probability). However, the timings of changes in trends for the individual proxy records show variations from the stack, indicating regional differences in the pattern of temperature change, particularly in the WAIS Divide record at the onset of the deglaciation; the Dome Fuji record at the deglaciation end; and the EDML record after 16 ka (EPICA Dronning Maud Land, where EPICA is the European Project for Ice Coring in Antarctica). In addition, two changes – one at 16 ka in the CO2 record and one after the ACR onset in three of the isotopic temperature records – do not have high-probability counterparts in the other record. The likely-variable phasing we identify testify to the complex nature of the mechanisms driving the carbon cycle and Antarctic temperature during the deglaciation.