C.5. Future Climate Scenarios

Estimation of the potential impacts of global warming should utilize several future climate scenarios, since the magnitude, timing and spatial details of global warming vary among climate models. Most published impacts studies were based on atmospheric General Circulation Model (GCM) doubled CO₂ radiative forcing equilibrium experiments with simple mixed-layer oceans. Doubled CO₂ radiative forcing (2 x CO₂) includes only about 50% actual CO₂ forcing with the balance arising from other greenhouse gases. More recent, transient experiments with coupled atmosphere-ocean GCMs have suggested a global average increase in temperature of about 1.0-3.5�C by the time of CO₂ doubling, estimated as 60-70 years from now (described in the IPCC Second Assessment Report, SAR; IPCC 1996, WG I, Section 6; Annex B). The most recent GCMs include sulfate aerosols in some experiments, which can cool the climate. The analysis presented here will rely both on the older 2 x CO₂ equilibrium GCM scenarios (described in the IPCC First Assessment Report, FAR; IPCC 1990, WG I, Section 3; Annex B), since most published analyses have relied on them, and on three new simulations, two from the Hadley Center (HADCM2GHG and HADCM2SUL; Johns et al., submitted; Mitchell et al., 1995; IPCC 1996, WG I, Sections 5, 6), and one from the Max Planck Institute for Meteorology (MPI-T106; Bengtsson, et al. 1995; Bengtsson, et al., 1996; IPCC 1996, WG I, Section 6), which have been made using coupled atmosphere-ocean GCMs and considering sulfate aerosol forcing.

To allow direct comparison with the previously completed VEMAP simulations over the conterminous U.S. (VEMAP Members, 1995), the same three equilibrium GCM scenarios were utilized for the global simulations: UKMO (Mitchell and Warrilow, 1987); GFDL-R30 (IPCC 1990, WG I, Section 3; IPCC 1990, WG I, Section 5); and OSU (Schlesinger and Zhao, 1989). The coarse grid from each model was interpolated to a 0.5� x 0.5�, lat.-long. grid. Scenarios were constructed by applying ratios ((2 x CO₂)/(1 x CO₂)) of all climate variables (except temperature) back to a baseline longterm average monthly climate dataset (Leemans and Cramer, 1991). Ratios were used to avoid negative numbers (e.g., negative precipitation), but were not allowed to exceed 5, to prevent unrealistic changes in regions with normally low rainfall. Temperature scenarios were calculated as a difference ((2 x CO₂) - (1 x CO₂)) and applied to the baseline dataset.

The newer GCM scenarios are extracted from transient GCM simulations wherein trace gases were allowed to increase gradually over a long period of years, allowing the climate to adjust while incorporating inherent lags in the ocean-atmosphere systems. In order to run the equilibrium vegetation models under the newer transient GCMs, a control climate is extracted as an average of either 30 years (Hadley Center) or 10 years (Max Planck Institute) of model output associated with present climate (e.g. 1961-1990). Likewise, a 30 or 10 year average is extracted from the time period approximating 2 x CO₂ forcing (e.g. 2070- 2099). These average climates are then used to drive the vegetation models. Note that because the vegetation models are equilibrium models, the results must be interpreted as indicating the potential vegetation, i.e., the climatically suitable vegetation. Time lags and transient responses of the vegetation to climate change are not considered here.

C.6. Interpretation of Biogeographic Model Simulations

Each of the ten IPCC regions was supplied with a set of MAPSS and BIOME3 output. Included were figures of vegetation distribution under current and future climate, vegetation density change (indexed by leaf area change), and runoff change. Also included were summary tables of the areas of the different biomes within each region under current and future climate, a change matrix indicating the area shifts from current biome type to other types, the areas within each biome expected to undergo an increase or decrease in vegetation density (change in LAI) and the areas within each biome expected to undergo an increase or decrease in annual runoff. These results were supplied for each vegetation model and for each GCM scenario. MAPSS and BIOME3 were both run under the Hadley Center scenarios; BIOME3 alone was run under the Max Planck Institute scenario; and, MAPSS alone was run under the older OSU, GFDL-R30 and UKMO scenarios. The Hadley and MPI simulations were run both with and without a direct CO₂ effect (applied in the ecological models); while, the OSU, GFDL-R30 and UKMO scenarios were only run with the direct CO₂ effects incorporated, in keeping with the VEMAP analyses.

Since the regional maps are of a much smaller extent and include quantitative information, the detailed interpretation will be left to the regions and the following discussion will only address general features of the simulations, particularly the differences between the older and newer GCMs and the MAPSS and BIOME3 intercomparisons. Although each region received the full set of figures, only a subset will be presented here. The MAPSS and BIOME3 results are sufficiently similar that the ranges presented in Tables C-1, C-2, C-3, C-4 and C-5 encompass the output from both models to indicate the full range of uncertainties within the scope of these experiments and models.

IPCC Homepage