GDGTs in the machine: Comparing Proxy Records of African Temperature Change to Holocene Climate Model Simulations via Proxy System Modeling

Current-generation climate models project that Africa will warm by up to 5 ℃ in the coming century, severely stressing African societies. To better constrain these model projections, this study compares lake sedimentary temperature proxies to model simulations of past climate change in Africa during the Holocene. Reconstructions from tropical African lakes derived from these lake sedimentary archives indicate 1-3° C of warming during the Mid-Holocene relative to the 20th century. Current generation paleoclimate models, however, do not replicate this proxy-reconstructed Mid-Holocene warming. This data-model mismatch could be due to a variety of factors, including model surface energy budgets or inaccurate representation of the 'Green Sahara.' To address this discrepancy, we employ a lake proxy system model to compare eight regional reconstructions of tropical African lake temperatures to surface air temperatures predicted by the Paleoclimate Modeling Intercomparison Project (PMIP3 and PMIP4). We examine the differences among model simulations in the PMIP ensemble focusing on heat budget terms to investigate the drivers of paleoclimate data-model mismatch. Our results suggest that PMIP models (in both versions 3 and 4) underestimate Mid-Holocene lake surface warming at our lake sites by approximately 2° C, potentially due to the joint impacts of boundary condition uncertainties in vegetation and Saharan dust. Finally, we extend this work by evaluating transient simulations from the TRACE, FAMOUS, and LOVECLIM projects spanning the full Holocene. We test the sensitivity of African temperatures to orbital insolation, greenhouse gasses, and ice-sheet dynamics in single-forcing transient simulations, and evaluate these patterns relative to our records. We elucidate the rates, amplitudes, and spatial heterogeneity of Holocene African temperature changes. We then discuss the implications of these findings for climate models' ability to simulate and predict future climate change in Africa.