Abstract/Summary

The CMIP5/PMIP3 mid-Holocene simulations show drier conditions in the Eurasian mid-continent and a significant increase in summer temperature; in contrast, paleoenvironmental data (including lake level, vegetation and isotope records, and aeolian deposits) and quantitative climate reconstructions show that the mid-continental extratropics were wetter than today and summers were cooler (Harrison et al., 2015). Eurasian mid-continental aridity and warming has been a persistent feature of model simulations, already present in atmosphere-only simulations (Yu & Harrison, 1996) and appearing more strongly in coupled ocean-atmosphere simulations (e.g. Braconnot et al., 2007b; Wohlfahrt et al., 2008; Harrison et al., 2015) and further exacerbated by vegetation feedback (Wohlfarht et al., 2004). The consistency among multiple lines of paleoenvironmental evidence makes it unlikely that the mismatch reflects misinterpretation of the data. Regional temperature biases in the CMIP5 20th century simulations have been linked to biases in surface energy and water balances, with over- or under-prediction of moisture fluxes and evapotranspiration leading to cold and warm temperature biases respectively (Mueller & Seneviratne, 2014). This suggests that discrepancies in the simulation of mid-Holocene climates might have a similar cause. In this paper, we investigate the processes involved in mid-continental climate changes in the CMIP5/PMIP3 simulations in order to identify the underlying cause of the mismatch with observations.