The Temporal Evolution of Changes in Carbon Storage in the Northern Permafrost Region Simulated by Carbon Cycle Models between 2010 and 2300: Implications for Atmospheric Carbon Dynamics

Thursday, 17 December 2015: 11:50
2004 (Moscone West)
Anthony David McGuire, U.S. Geological Survey, University of Alaska Fairbanks, Fairbanks, AK, United States, David M Lawrence, National Center for Atmospheric Research, Boulder, CO, United States, Eleanor Burke, Met Office Hadley Centre, Exeter, United Kingdom, Guangsheng Chen, Oak Ridge National Laboratory, Oak Ridge, TN, United States, Elchin E Jafarov, Institute of Arctic and Alpine Research, Boulder, CO, United States, Charles Koven, Lawrence Berkeley National Laboratory, Berkeley, CA, United States, Andrew H MacDougall, University of Victoria, Victoria, BC, Canada, Dmitry Nicolsky, University of Alaska Fairbanks, Fairbanks, AK, United States, Shushi Peng, Peking University, Beijing, China and Annette Rinke, Alfred Wegener Institute Helmholtz-Center for Polar and Marine Research Potsdam, Potsdam, Germany
We conducted an assessment of changes in permafrost area and carbon storage simulated by 8 process-based models between 2010 and 2300. The models participating in this comparison were those that had joined the model integration team of the Vulnerability of Permafrost Carbon Network (see http://www.permafrostcarbon.org/). Each of the models in this comparison conducted simulations over the permafrost land region in the Northern Hemisphere driven by CCSM4-simulated climate for RCP 4.5 and 8.5 scenarios. Among the models, the area of permafrost (defined as the area for which active layer thickness was less than 3 m) in 2010 ranged between 8 and 19 million km2. Between 2100 and 2300, models indicated the loss of permafrost area between 3 and 5 million km2 for RCP 4.5 and between 6 and 16 million km2 for RCP 8.5. Among the models, the density of soil carbon storage in 2010 ranged between 10 and 45 thousand g C m-2; models that explicitly represented carbon with depth had estimates greater than 32 thousand g C m-2. For the RCP 4.5 scenario, mean cumulative change in soil carbon between 2010 and 2300 was a gain of 10 Pg C (range: loss of 67 to gain of 70 Pg C). For the RCP 8.5 scenario, the mean cumulative change in soil carbon was between 1960 and 2300 was a loss of 256 Pg C (range: losses of 7 to 652 Pg C). Gains in vegetation carbon negated losses in the RCP 4.5 simulations for all but one of the models (mean change in total ecosystem carbon: 60 Pg C, range: loss of 14 Pg C gain of 244 Pg C), but only for two of the RCP 8.5 simulations (mean: 148 Pg C, range: loss of 641 to gain of 167 Pg C). For the RCP simulations that lost carbon between 2010 and 2300, substantial losses of carbon did not occur until after 2100. These results suggest that the permafrost carbon feedback would not have substantial consequences until after 2100, and that effective mitigation efforts during this century have the potential to prevent the negative consequences of the permafrost carbon feedback.