|The Regional Impacts of Climate Change|
|Other reports in this collection|
If the direct effects of CO2 are minimal and the future scenarios are relatively warm, decreases in LAI could occur over very large forested areas, ranging up to nearly 2/3 or more of the areas of boreal, temperate and tropical forests (Table C-2). By contrast, if the direct effects of CO2 are strong and scenarios are not too warm, then all forest vegetation zones could experience increased biomass over as much as 2/3 or more of their areas (Table C-3). More likely, the responses will be intermediate with large regional contrasts, decreases in vegetation density in some areas, increases in others. Even though these are equilibrium simulations, a simulated decline in LAI generally implies a less favorable water balance and a loss of vegetation density. These losses imply a process of loss over some time period. We can only draw inferences about how rapidly such losses would occur, based on the simulated amount of loss. The regions that could experience declining LAI (Figure C-6, Figure C-7, Figure C-8 and Figure C-9), would exhibit spatial gradients in response from mild decline grading into potentially catastrophic dieback. All reaches along the decline gradients would experience drought stress, which could trigger other responses, such as pest infestations and fire. Following disturbance by drought, infestation or pests, new vegetation, either of the same or of a different type would grow, but to a lower density.
Including both equilibrium and 'transient' scenarios, MAPSSwas run under four different scenarios (not counting the sulfate scenario, HADSUL). These range in global temperature increase (delta T) at the time of 2 x CO2 from 1.7 (HADGHG) to 5.2°C (UKMO) (Annex B). In general, the areas of forest decline within individual biomes (incorporating a direct CO2 effect) increase linearly with increasing delta T in the temperate and boreal forests; while, the areas of increased forest density decrease with increasing delta T. Tropical forests exhibit a similar pattern across the three FAR scenarios, but under the cooler HADGHG scenario show a large decline as simulated by MAPSS. By contrast, BIOME3, under the HADGHG scenario, shows almost no change in tropical forest density. Interestingly, adjacent tropical savannas increase in density in both ecological models under the HADGHG scenario.
The newer climate scenarios (IPCC 1996, WG I, Section 6), extracted from transient GCM simulations, are as a group quite different from the older, equilibrium scenarios (IPCC 1990, WG I, Section 3), in terms of the simulated ecological responses that these scenarios produce. All of the older scenarios produce large regions showing LAI declines (especially in temperate to high latitudes), as well as gains, even when the direct effects of CO2 are included (MAPSS simulations, Figure C-6, OSU and UKMO scenarios not shown). By contrast, under the newer scenarios, if a direct CO2 effect is assumed, then there are very few regions with declines in LAI, as simulated by both MAPSS and BIOME3 (Figures C-7, C-8); rather, most of the world is simulated with an increased LAI. Actual increases in LAI could be limited by nitrogen availability in some areas, although elevated soil temperatures could increase decomposition, releasing more nitrogen (McGuire et al., 1995; VEMAP Members, 1995). The first-order differences between the older and newer scenarios are likely due to the smaller global temperature increases in the newer climate scenarios, which came from GCMs that had not attained their full equilibrium temperature changes.
The incorporation of sulfate aerosols produced a cooling effect in the HADCM2SUL run compared to the HADCM2GHG run, which lacked the sulfate forcing (GHG runs are not shown). The vegetation response to the sulfate forcing is observable in the model output from both MAPSS and BIOME3, but is relatively small compared to the differences between the newer and older climate scenarios. The newer scenarios produce widespread enhanced vegetation growth, even without the sulfate effect, if direct CO2 effects are included and widespread decline if the CO2 effects are excluded. The presence of the sulfate-induced cooling produces a much smaller amplitude effect on the vegetation than does the presence or absence of the direct effects of elevated CO2 on water-use-efficiency.
Changes in annual runoff (Figure C-10) were mapped
for all scenarios from both MAPSS and BIOME3. The changes in runoff are more
stable among the different climate scenarios than are the simulated changes
in LAI. The relative stability of simulated runoff change may reflect that runoff
is a passive drainage process; whereas, evapotranspiration is a biological process
and a function of the product of LAI and stomatal conductance. If stomatal conductance
is reduced, e.g., via a direct CO2 effect, LAI will compensate by increasing
and runoff will show little change (Neilson and Marks, 1994). Some of the obvious
differences between MAPSS and BIOME3 can be attributed to structural differences
in the models. BIOME3 calculates water balance daily, even though all inputs
are monthly; whereas, MAPSS calculates water balance monthly. This difference
alone could be causing the more extreme responsiveness of MAPSS, which shows
both larger runoff increases and larger losses in different regions. On the
other hand, MAPSS uses a 3-layer soil with roots only in the top two layers;
while BIOME3 uses a 2-layer soil with roots in both layers. The third layer
in MAPSS provides a consistent base flow and might explain why MAPSS produces
runoff in some drier regions, such as the western U.S., while BIOME3 does not.
The hydrology models in both MAPSS and BIOME3, although process-based, are considered
prototypes for eventual replacement by more elaborate models (see for example,
the PILPS model intercomparison study; Love and Henderson-Sellers, 1994).
In general, MAPSS and BIOME3 produce similar regional patterns in the estimated changes in runoff. Although the magnitude of the changes are different, there are broad similarities in the sign of the change (but, clearly not in all regions). The largest area of regional difference between the two models is in interior Eurasia (Figure C-10).
Runoff generally increases in the Tundra, due to higher temperatures, more precipitation and more melting (Tables C-4, C-5). It decreases in the Taiga/Tundra due to encroachment of high-density boreal forest into low density vegetation (hence, higher transpiration). Runoff results are varied in the temperate forests, but Temperate Mixed forests tend to present a higher likelihood of reduced runoff over large areas (range 51% to 88% of the area under all scenarios) than of increased runoff (range 11% to 47% of the area under all scenarios, Tables C-4, C-5). Even the most benign scenarios indicate a minimum of 51% of the area of the world's temperate evergreen forests could experience a runoff decline; whereas, a maximum of 47% of the area would experience increased runoff. Temperate Evergreen Forests exhibit a greater likelihood of increased runoff over large areas (range 29% to 87% of the area under all scenarios) than decreased runoff (range 11% to 68%), but the overlap in these increase and decrease ranges indicates the degree of uncertainty in the simulations. However, much of the increased runoff in the Temperate Evergreen forested areas is due to increased winter runoff, which is not necessarily available for use by ecosystems, irrigation or domestic purposes. Runoff from tropical forest areas could either increase or decrease over large areas, depending largely on the importance of the direct CO2 effects. Runoff from drier vegetation types is regionally variable and exhibits both increases and decreases, depending on the direct CO2 effects and regional rainfall patterns.
Other reports in this collection