Likelihood of a decrease in runoff (a), an increase in runoff (b) and a severe change in ecosystems (c) for selected ΔTg levels

Figure 2. Likelihood of a decrease in runoff (a), an increase in runoff (b) and a severe change in ecosystems (c) for selected ΔTg levels. (a) and (b) show whether the simulated decrease (increase) in average annual runoff exceeds present (1980–2009) standard deviation, or whether monthly runoff is >10% more frequently below (above) its present median. Areas with presently <10 mm yr−1 are masked out. The likelihoods are derived from the 19 climate change patterns. See figures S1–S4 (available at stacks.iop.org/ERL/8/034032/mmedia) in the supplement for all eight ΔTg levels.

Abstract

This modelling study demonstrates at what level of global mean temperature rise (ΔTg) regions will be exposed to significant decreases of freshwater availability and changes to terrestrial ecosystems. Projections are based on a new, consistent set of 152 climate scenarios (eight ΔTg trajectories reaching 1.5–5 ° C above pre-industrial levels by 2100, each scaled with spatial patterns from 19 general circulation models). The results suggest that already at a ΔTg of 2 ° C and mainly in the subtropics, higher water scarcity would occur in >50% out of the 19 climate scenarios. Substantial biogeochemical and vegetation structural changes would also occur at 2 ° C, but mainly in subpolar and semiarid ecosystems. Other regions would be affected at higher ΔTg levels, with lower intensity or with lower confidence. In total, mean global warming levels of 2 ° C, 3.5 ° C and 5 ° C are simulated to expose an additional 8%, 11% and 13% of the world population to new or aggravated water scarcity, respectively, with >50% confidence (while ~1.3 billion people already live in water-scarce regions). Concurrently, substantial habitat transformations would occur in biogeographic regions that contain 1% (in zones affected at 2 ° C), 10% (3.5 ° C) and 74% (5 ° C) of present endemism-weighted vascular plant species, respectively. The results suggest nonlinear growth of impacts along with ΔTg and highlight regional disparities in impact magnitudes and critical ΔTg levels.