Vulnerability of Permafrost Carbon: Process-Based Model Analysis 1960-2009

Presentation given by David McGuire at the 2013 American Geophysical Union Fall Meeting.

Abstract: We conducted a retrospective (1960 - 2009) comparison of how large-scale models represent permafrost carbon dynamics. The models participating in this comparison were those that had joined the model integration team of the Vulnerability of Permafrost Carbon Research Coordination Network (see Each of the 9 models in this comparison conducted simulations over the permafrost land region in the Northern Hemisphere. Among the models, the area of permafrost (defined as the area for which active layer thickness was less than 3 m) ranged between 7.4 and 28.5 million km2 and the density of soil carbon storage ranged an order of magnitude between 9.9 and 85.7 thousand g C m-2. Between 1960 and 2009, models generally indicated loss of permafrost area that ranged between 9.4 thousand km2 and 2.8 million km2. Although the permafrost area decreased, models simulated gains in soil carbon storage that ranged from a gain of 24 g C m-2 to a gain of 1032 g C m-2. All models indicated that both net primary production (NPP) and heterotrophic respiration (RH) increased from 1960, with a mean increase of NPP that was approximately 0.23 g C m-2 yr-1 greater than the increase in RH. However, there are indications among the models that the NPP anomalies are decelerating in magnitude by the end of the analysis period, while the RH anomalies are accelerating. Some of the models are clearly showing a deceleration in the accumulation of soil carbon during the last fifty years. These results suggest that simulated RH may generally overtake simulated NPP in applications of these models driven by future climate projections, a response that would result in net losses of carbon from permafrost zone soils.