Modelling framework for asynchronous land-atmosphere coupling using NASA GISS ModelE (NASA-GISS E2.1) and LPJ-LMfire (v1.4.0): design, application and evaluation for the 2.5 ka period

Abstract. While paleoclimate simulations have been a priority for Earth system modelers over the past three decades, little attention has been paid to the period between the mid-Holocene (6 ka) and the Last Millennium, although this is an important period for the emergence of complex societies. Here, we consider the climate of 2500 BP (before present or 550 BCE), a period when compared to late preindustrial time, greenhouse gas concentrations were slightly lower, and orbital forcing led to a stronger seasonal cycle in high latitude insolation. To capture the influence of land cover on climate, we asynchronously coupled the NASA GISS ModelE Earth system model with the LPJ-LMfire dynamic global vegetation model. We simulated global climate and assessed our results in the context of independent paleoclimate reconstructions. We also explored a set of combinations of model performance parameters (bias and variability) and demonstrated their importance for the asynchronous coupling framework. The asynchronously coupled model system shows strong vegetation-albedo feedback on climate and is comparatively more sensitive to the bias correction than the internal model variability and green Sahara conditions. In the absence of a bias correction, while driving LPJ-LMfire in the coupling process, ModelE drifts towards colder conditions in the high latitudes of the Northern Hemisphere in response to land cover simulated by LPJ-LMfire. A regional precipitation response is also prominent in the various combinations of the coupled model system, with a substantial intensification of the Summer Indian Monsoon and a drying pattern over Europe. Evaluation of the simulated climate against reconstructions of temperature from multiple proxies and the isotopic composition of precipitation (δ18Op) from speleothems demonstrated the skill of ModelE in simulating past climate. A regional analysis of the simulated vegetation-climate response further confirmed the validity of this approach. The NASA GISS ModelE was found to be particularly sensitive to the representation of shrubs and this land cover type requires particular attention as a potentially important driver of climate in regions where shrubs are abundant. Our results further demonstrate the importance of bias correction in coupled paleoclimate simulations. This study presents a generalized framework for incorporating biogeophysical responses into climate models without dynamic vegetation, for simulating past climates, in line with the recommendations of the Paleoclimate Modelling Intercomparison Project (PMIP).