A model-data comparison of the Last Glacial Maximum surface temperature changes

Abstract. Over the Last Glacial Maximum (LGM, ~ 21 ka BP), the presence of vast Northern Hemisphere ice-sheets caused abrupt changes in surface topography and background climatic state. While the ice-sheet extent is well known, several conflicting ice-sheet topography reconstructions suggest that there is uncertainty in this boundary condition. The terrestrial and sea surface temperature (SST) of the LGM as simulated with six different Laurentide Ice Sheet (LIS) reconstructions in a fully coupled Earth System Model (COSMOS) have been compared with the subfossil pollen and plant macrofossil based and marine temperature proxies reconstruction. The terrestrial reconstruction shows a similar pattern and in good agreement with model data. The SST proxy dataset comprises a global compilation of planktonic foraminifera, diatoms, radiolarian, dinocyst, alkenones and planktonic foraminifera Mg / Ca-derived SST estimates. Significant mismatches between modeled and reconstructed SST have been observed. Among the six LIS reconstructions, Tarasov’s LIS reconstruction shows the highest correlation with reconstructed terrestrial and SST. In the case of radiolarian, Mg / Ca, diatoms and foraminifera show a positive correlation while dinocyst and alkenones show very low and negative correlation with the model. Dinocyst-based SST records are much warmer than reconstructed by other proxies as well as Pre-industrial (PI) temperature. However, there are large discrepancies between model temperatures and temperature recorded by different proxies. Eight different PMIP3 models also compared with temperature proxies reconstruction which show mismatches with the proxy records might be due to misinterpreted and/or biased proxy records. Therefore, it has been speculated that considering different habitat depths and growing seasons of the planktonic organisms used for SST reconstruction could provide a better agreement of proxy data with model results on a regional scale. Moreover, it can reduce model-data misfits. It is found that shifting in the habitat depth and living season can remove parts of the observed model-data mismatches in SST anomalies.