Quantitative paleoclimatic reconstructions for the Late Quaternary of southern South America based on calibration of modern pollen and climate relationships

Three approaches to deriving quantitative estimates of past temperature and precipitation (multiple regression, response surfaces, and the method of modern analog) were evaluated using a modern pollen and climate data set from southern South America. The results of this analysis show that, under conditions of sparse climate data, response surfaces are the most robust approach. The three calibrated data sets were then applied to four pollen records, covering the last glacial - interglacial transition, to derive quantitative estimates of changes in temperature and precipitation for the South American mid-latitudes (41{dollar}spcirc{dollar}). Although there are variations among the records, there is generally good agreement, indicating largely consistent regional scale change. Comparisons of the quantitative climate reconstructions with qualitative records of paleoenvironmental conditions in the region show good agreement.; Data from the Lake District of Chile and Argentina were then integrated with quantitative paleotemperature records from the equatorial region of South America, and with temperature estimates from the Byrd (Antarctica) {dollar}deltasp{lcub}18{rcub}{dollar}O record. This paleoenvironmental data set was compared to general circulation model (GCM) scenarios of climate change. Three different sets of model output from the GIST-NESS GCM for the LGM, each with different a different set of sea surface temperature (SST) values were compared. The model run using calculated values for SST based on an ocean heat convergence (OHC) and ocean heat transport (OHT) model reproduced the most accurate surface temperature values for the equatorial and polar regions. However, this model did not perform well in the mid-latitudes {dollar}(30spcirc{dollar}-60{dollar}spcirc{dollar}S). Model output from the NCAR community climate model (CCM1) for time slices spanning from the LGM to 6000 years ago were then compared to the paleoenvironmental data set. This comparison shows that the model's linear response to changing boundary conditions is not an accurate reconstruction of the pattern of climate change that has occurred since the LGM. The paleoenvironmental data show that the onset of warming occurred first at the high and mid-latitudes, preceding equatorial warming by as much as 7000 years. The paleoenvironmental data also show that the continuous warming shown in the model output is inaccurate. Warming occurred abruptly, and reversals are recorded at all latitudes. In the polar region, temperatures jumped 5{dollar}spcirc{dollar}C from 21 Ka to 16 Ka, and had reached modern conditions by 14 Ka. At the mid-latitudes, temperatures transitioned from the full glacial minimum to warmer than today between 21 Ka and 16 Ka, while at the equator, temperatures did not start to warm until ca. 14 Ka.