Amplification of Warming at Tropical High Altitudes: Synthesis of East African Lacustrine and Moraine Archives with Paleoclimate Model Data

Tuesday, 16 December 2014: 4:15 PM
Shannon E Loomis, University of Texas at Austin, Austin, TX, United States, James M Russell, Brown University, Providence, RI, United States and Carrie Morrill, University of Colorado at Boulder, Boulder, CO, United States
The amplitude of tropical paleotemperature variability on glacial/interglacial time scales has been hotly debated for decades, as sea surface temperature (SST) reconstructions from the tropics have repeatedly shown less warming than reconstructions from high elevations from the Last Glacial Maximum (LGM) to present. This debate is complicated by the fact that SST and high elevation reconstructions rely on different proxies, which inherently contain different errors and assumptions. Here we present a new equatorial paleotemperature record from Lake Rutundu, Mt. Kenya (3081 m above sea level) based on branched glycerol dialkyl glycerol tetraethers (GDGTs) and compare it to previously published East African GDGT temperature reconstructions at different elevations to assess the timing and amplitude of temperature changes with altitude. We find that the overall amplitude of warming in the Lake Rutundu record is 6.8±1.0°C from the LGM to the present, with mid-Holocene temperatures 1.6±0.9°C warmer than modern.

By combining our Lake Rutundu record with three previously published GDGT temperature records from different elevations in East Africa (Sacred Lake, Lake Tanganyika, and Lake Malawi), we can calculate changes in lapse rates and freezing level heights (FLHs) through time. We find that lapse rates were steeper at the LGM (-7.4ºC/km) compared to the modern (-6.1ºC/km), likely due to widespread regional aridity at the LGM. FLHs were nearly 1000 m lower than they are today, which is supported by regional equilibrium line altitude (ELA) reconstructions from East African moraines. Patterns of altitudinal temperature variability and lapse rate changes reflect those from the TraCE-21 paleoclimate model, but amplitudes of model data variability are much smaller.