Strengths and challenges for transient Mid- to Late Holocene simulations with dynamical vegetation

• Published in: Climate of the Past
• Main Authors: Braconnot, Pascale, Zhu, Dan, Marti, Olivier, Servonnat, Jérôme
• Format: Article in Journal/Newspaper
• Language: English
• Published: Copernicus Publications 2019
• Subjects: article; Verlagsveröffentlichung; Sea ice
• Online Access: https://doi.org/10.5194/cp-15-997-2019; https://noa.gwlb.de/receive/cop_mods_00001683; https://noa.gwlb.de/servlets/MCRFileNodeServlet/cop_derivate_00001642/cp-15-997-2019.pdf; https://cp.copernicus.org/articles/15/997/2019/cp-15-997-2019.pdf

We present the first simulation of the last 6000 years with a version of the IPSL Earth system model that includes interactive dynamical vegetation and carbon cycle. It is discussed in the light of a set of Mid-Holocene and preindustrial simulations performed to set up the model version and to initialize the dynamical vegetation. These sensitivity experiments remind us that model quality or realism is not only a function of model parameterizations and tunings but also of experimental setup. The transient simulations shows that the long-term trends in temperature and precipitation have a similar shape to the insolation forcing, except at the Equator, at high latitudes, and south of 40∘ S. In these regions cloud cover, sea ice, snow, or ocean heat content feedbacks lead to smaller or opposite temperature responses. The long-term trend in tree line in the Northern Hemisphere is reproduced and starts earlier than the southward shift in vegetation over the Sahel. Despite little change in forest cover over Eurasia, a long-term change in forest composition is simulated, including large centennial variability. The rapid increase in atmospheric CO2 in the last centuries of the simulation enhances tree growth and counteracts the long-term trends induced by Holocene insolation in the Northern Hemisphere and amplifies it in the Southern Hemisphere. We also highlight some limits in the evaluation of such a simulation resulting from model climate–vegetation biases, the difficulty of fully assessing the result for preindustrial or modern conditions that are affected by land use, and the possibility of multi-vegetation states under modern conditions.