Research Projects

We investigate the drivers of simulated model temperatures in tropical Africa during the mid-Holocene, motivated by the 'Holocene temperature conundrum' wherein geological and bio/geochemical records of past temperature suggest there was a thermal maximum 6,000 years ago that models tend not to simulate. We performed an energy budget decomposition analysis on models from the Paleoclimate Modeling Intercomparison Project (PMIP3 and 4) that allows us to quantify the contribution of different energy balance terms to overall temperature change, and found that changes in local hydroclimate were the primary drivers of cooling. This suggests that PMIP models may overestimate cooling in response to the strengthened African monsoon. We also investigated two 'Green Sahara' experiments that either prescribe vegetation to the Sahara or suppress dust fluxes into the atmosphere, and found that these experiments bring model simulated temperatures closer to proxy reconstructions.

This manuscript investigates proxy/model disagreement from the proxy side, using a proxy system model to compare a suite of lake surface temperature reconstructions from glycerol dialkyl glycerol tetraethers (GDGTs) to climate model simulations of the Last Glacial Maximum and mid-Holocene. We find that the proxy records have significantly more temperature variability than the PMIP models, recording temperatures 2 degrees cooler than the models during the LGM and 2 degrees warmer during the mid-Holocene. We then quantify the impact of different sources of proxy error on the overall temperature discrepancy, including whether the GDGTs record seasonal temperatures or the relationship between lake and air temperature is non-stationary over time. We determine that proxy system error can account for a significant portion (but not all) of the temperature disagreement during the mid-Holocene, and very little of the disagreement during the Last Glacial Maximum. 

For the last chapter of my dissertation, I plan to compare the same suite of GDGT lake surface temperature proxies to a group of transient model simulations of the last 21,000 years to see if models can capture the rates and spatial patterns of temperature change, using principal component analysis (PCA).  So far, we have found that the proxy reconstructions record faster rates of temperature change than the transient models during the entire Deglaciation and Holocene, but the transient models do capture the spatial pattern of warming well. We also plan to look at single forcing experiments to determine what signals in the models most closely match proxy reconstructions of temperature change.