Abstract
As more modeling groups employ fully coupled Earth System Models that simulate climate and biogeochemical fluxes simultaneously, it is critical that we understand how regional errors in temperature and precipitation produced by atmosphere-ocean GCMs may alter our understanding of vegetation dynamics, and subsequent changes in terrestrial carbon stocks, irrespective of biogeochemical or biophysical feedbacks. To test the influence of regional errors on projected vegetation and terrestrial carbon changes, we performed two sets of calculations to mimic 1) uncorrected coupled and 2) bias-corrected off-line simulations of a dynamic vegetation model. We used historical (Climate of the 20th Century) and future (A1B) output from 18 of the IPCC AR4 GCMs as input for a simplified Köppen classification scheme associated with carbon density values derived from the Olson carbon database. We found that although global changes in above-ground terrestrial carbon century differed little between the two methods by the end of the 21st century (2079-99), the location of sources and sinks that influence the global carbon balance are distinct. In “bias-corrected” simulations, the direction of change in global carbon stocks is dependent on whether the model projection promotes conversion of tropical forest to savanna in South America. In “uncorrected” vegetation simulations, however, models that gain and loose carbon differ primarily in the strength of the carbon sink associated with the transition of Northern Hemisphere tundra to boreal forest. These distinctions are due primarily to how closely the modern (1979-99) climate simulated by a given model reproduces the observed distribution of Köppen types in the “uncorrected” simulation, as the relative changes in climate were the same for the two experiments. This source of disparity should be taken into account when comparing studies using these two methods to future changes in climate. Of particular importance are changes in vegetation in the high- versus low-latitudes and the associated with differences in their primary biophysical feedbacks to climate that may suggest dissimilar adaptation and/or mitigation strategies.
spacer