The past 21,000 years of tropical African temperature change in GDGT lake sediment proxies and equilibrium and transient model simulations

Paleoclimate records help evaluate climate model performance in different climate states to constrain predictions of future climate change. There is significant disagreement between proxy records of tropical African temperature and climate model simulations, raising the possibility that these models under- or over-estimate future warming. We compare glycerol dialkyl glycerol tetraether (GDGT) records from eight lakes in eastern tropical Africa to simulations from equilibrium and transient climate model experiments. We find there is significant disagreement, particularly at the mid-Holocene and Last Glacial Maximum, which could arise from errors within the proxies, the models, or both. We examine potential sources of error in both the models and the proxies to quantify the extent to which each contributes to the temperature discrepancy. For the proxies, we utilize a proxy system model (PSM) to quantify potential non-stationarity and seasonality in the proxy records; we also explicitly simulate and analyze the impact of calibration uncertainty on GDGT-reconstructed temperature. The combined potential sources of error account for a significant portion (~2oC) of the proxy/model disagreement. For the models, a surface energy budget decomposition is used to determine what processes and feedbacks the models do not simulate in the warmer mid-Holocene. Changes in hydroclimate relating to the ‘Green Sahara’ and strengthened West African Monsoon may contribute to warming in the proxies that is not simulated by the models. We also analyze transient model experiments and find that they record a slower rate of temperature change than the proxies, but the simulations do capture the spatial patterns of warming in Africa over the past 22 kyr. This comprehensive paleoclimate data-model comparison for Africa indicates that GCMs may be at least partially responsible for proxy/model temperature disagreement, highlighting the necessity of improvements to model physics and/or boundary conditions for simulations of this region.