Modelling carbon responses of tundra ecosystems to historical and projected climate: a comparison of a plot and a global-scale ecosystem model to identify process-based uncertainties

We are developing a process-based modelling approach to investigate how car bon (C) storage of tundra across the entire Arctic will respond to projecte d climate change. To implement the approach, the processes that are least u nderstood, and thus have the most uncertainty, need to be identified and st udied. In this paper, we identified a key uncertainty by comparing the resp onses of C storage in tussock tundra at one site between the simulations of two models - one a global-scale ecosystem model (Terrestrial Ecosystem Mod el, TEM) and one a plot-scale ecosystem model (General Ecosystem Model, GEM ). The simulations spanned the historical period (1921-94) and the projecte d period (1995-2100). In the historical period, the model simulations of ne t primary production (NPP) differed in their sensitivity to variability in climate. However, the long-term changes in C storage were similar in both s imulations, because the dynamics of heterotrophic respiration (R-H) were si milar in both models. In contrast, the responses of C storage in the two mo del simulations diverged during the projected period. In the GEM simulation for this period, increases in R-H tracked increases in NPP, whereas in the TEM simulation increases in RH lagged increases in NPP. We were able to ma ke the long-term C dynamics of the two simulations agree by parameterizing TEM to the fast soil C pools of GEM. We concluded that the differences betw een the long-term C dynamics of the two simulations lay in modelling the ro le of the recalcitrant soil C. These differences, which reflect an incomple te understanding of soil processes, lead to quite different projections of the response of pan-Arctic C storage to global change. For example, the ref erence parameterization of TEM resulted in an estimate of cumulative C stor age of 2032 gC m(-2) for moist tundra north of 50 degreesN, which was subst antially higher than the 463 g C m(-2) estimated for a parameterization of fast soil C dynamics. This uncertainty in the depiction of the role of reca lcitrant soil C in long-term ecosystem C dynamics resulted from our incompl ete understanding of controls over C and N transformations in Arctic soils. Mechanistic studies of these issues are needed to improve our ability to m odel the response of Arctic ecosystems to global change.