Simulated Terrestrial Warming at the Last Interglacial Due to Vegetation Change

Abstract Paleo‐proxy records suggest that vegetation at the Last Interglacial (LIG) was different from pre‐industrial (PI), including a greener Sahara and a northward expanded boreal forest at high northern latitudes. Not accounting for these differences may have contributed to an underestimation of temperature and precipitation changes in paleoclimate simulations. Here, we use the LPJ‐GUESS dynamic vegetation model to generate a static vegetation map for the LIG as a boundary condition for the ACCESS‐ESM1.5 model to assess the impact of vegetation changes on climate. Compared to PI, bare ground fractions decrease globally, resulting in a more vegetated state, particularly in northern Africa and at high northern latitudes, where boreal forests expand and tundra shifts northward. As a result, annual mean global surface temperature increases by 0.2C compared to an initial LIG experiment that uses fixed PI vegetation cover, with warming mostly simulated over land due to albedo changes. The strongest warming (C) is located over the western Indian subcontinent, the far eastern region of Siberia, and northeastern North America, all associated with smaller bare ground fractions and albedo. Winds over land are weakened due to higher roughness, though the large‐scale global circulation patterns remain unchanged. Precipitation and monsoon changes are minor. The vegetation change implemented here leads to an annual mean warming of 0.5C over land in the northern hemisphere, which is relatively minor compared to LIG simulations that employ fully coupled dynamic global vegetation models, highlighting the importance of positive climate‐vegetation feedbacks that are not captured within our framework.