|Working Group II: Impacts, Adaptation and Vulnerability|
|Other reports in this collection|
12.1.5. Climate Scenarios Used in Regional Studies
Most recent impact studies for Australia and New Zealand have been based on scenarios released by the Commonwealth Scientific and Industrial Research Organisation (CSIRO, 1996a) or the National Institute of Water and Atmosphere (NIWARenwick et al., 1998b). The CSIRO (1996a) scenarios included two sets of rainfall scenarios that are based on results from equilibrium slab-ocean general circulation model (GCM) and transient coupled ocean-atmosphere GCM (AOGCM) simulations. Some impact and adaptation studies since RICC have used results from a 140-year simulation over the whole region that uses the CSIRO regional climate model (RCM) at 125-km resolution nested in the CSIRO Mark 2 coupled GCM run in transient mode from 1961 to 2100, as well as similar simulations at 60-km resolution over eastern Australia. Figure 12-1 (from Whetton et al., 2001) summarizes results of slab-ocean and coupled AOGCM simulations of rainfall changes for various subregions of Australia in summer (December-January-February, DJF) and winter (JJA), respectively.
Figures 12-1a and 12-1b clearly show a variety of estimates for different model simulations; nevertheless, there is a tendency toward more decreases in winter rainfall in the southwest and the central eastern coast of Australia in the coupled model results than in the slab-ocean model results and a similar tendency for more agreement between coupled model results on less rainfall in the northwest and central west in summer.
In line with WGI conclusions, more recent scenarios, such as those of Hulme and Sheard (1999) and Carter et al. (2000) rely exclusively on coupled models; they also are based on the newer emissions scenarios from the IPCC's Special Report on Emissions Scenarios (SRES) rather than the IS92 scenarios described in IPCC (1996). This leads to some changes in the regional scenarios. According to Whetton (1999), the CSIRO (1996a) scenarios give warmings of 0.3-1.4°C in 2030 and 0.6-3.8°C in 2070 relative to 1990, compared to estimates by Hulme and Sheard (1999) of 0.8-3.9°C by the 2050s and 1.0-5.9°C in the 2080s, relative to the 1961-1990 averages. As discussed by Whetton (1999), both sets of scenarios use results from several coupled models, but the use of the SRES emissions scenarios leads to greater warmings in Hulme and Sheard (1999) than those that are based on the IS92 emission scenarios in CSIRO (1996). Nevertheless, the Hulme and Sheard (1999) results are preliminary in that they use scaled results from non-SRES simulations, rather than actual GCM simulations with SRES emissions.
Rainfall scenarios that are based on the coupled model results in CSIRO (1996a) show decreases of 0-8% in 2030 and 0-20% in 2070, except in southern Victoria and Tasmania in winter and eastern Australia in summer, where rainfall changes by -4 to +4% in 2030 and -10 to +10% in 2070. The Hulme and Sheard (1999) scenarios suggest that annual precipitation averaged over either northern or southern Australia is likely to change by -25 to +5%. Larger decreases are indicated for the 2080s and for some specific locations (e.g., a 50% decrease in parts of the southwest of western Australia in winter). This last finding suggests that even though the rainfall decrease in this region in the late 20th century was almost certainly dominated by natural variability (Smith et al., 2000), a decreasing trend resulting from the enhanced greenhouse effect may dominate in this region by the mid- to late 21st century.
It is important to note that changing from joint use of equilibrium slab-ocean GCM and transient coupled AOGCM results to exclusive reliance on coupled model results in the Australian scenarios leads to a marked narrowing of the uncertainty range in rainfall changes predicted for southern Australia, with a tendency to more negative changes on the mainland (see Figure 12-1; Hulme and Sheard, 1999; Carter et al., 2000). This means that the results of impact studies that used the wider range of rainfall scenarios, as in the CSIRO (1996a) scenarios (e.g., Schreider et al., 1996), should be reinterpreted to focus on the drier end of the previous range. This is consistent with more recent studies by Kothavala (1999) and Arnell (1999).
To summarize the rainfall results, drier conditions are anticipated for most of Australia over the 21st century. However, consistent with conclusions in WGI, an increase in heavy rainfall also is projected, even in regions with small decreases in mean rainfall. This is a result of a shift in the frequency distribution of daily rainfall toward fewer light and moderate events and more heavy events. This could lead to more droughts and more floods.
Recent Australian impact studies have tended to use regional scenarios of temperature and rainfall changes per degree of global warming, based on the CSIRO RCM transient simulations, scaled to the range of uncertainty of the global warming derived from the IPCC Second Assessment Report (SAR) range of scenarios. For example, Hennessy et al. (1998) gives scenarios for six regions of New South Wales (NSW), based on a 60-km resolution simulation. Ranges are given for estimated changes in maximum and minimum temperatures, summer days over 35°C, winter days below 0°C, seasonal mean rainfall changes, and numbers of extremely wet or dry seasons per decade. Statistical downscaling (discussed in detail in et al. 3) has not been used extensively in Australia apart from work by Charles et al. ( 1999) on the southwest of western Australia. It should be noted that in some cases, RCM results may change the sign of rainfall changes derived from coarser resolution GCMs, because of local topographic and other effects (Whetton et al., 2001).
New Zealand scenarios reported in RICC (Basher et al., 1998) are based on statistical downscaling from equilibrium GCM runs with CO2 held constant at twice its present concentration. Renwick et al. (1998b, 2000) also have downscaled equilibrium GCM simulations over New Zealand, using a nested RCM.
A key factor in rainfall scenarios for New Zealand is the strength of the mid-latitude westerlies because of the strong orographic influence of the backbone mountain ranges lying across this flow (Wratt et al., 1996). Earlier equilibrium slab-ocean GCM runs (applicable only long after stabilization of climate change) and the regional simulations obtained by downscaling them through statistical and nested modeling techniques predict a weakening of the westerlies across New Zealand, particularly in winter. As a result, downscaled equilibrium model runs predict that winter precipitation will remain constant or decrease slightly in the west but increase in Otago and Southland (Whetton et al., 1996a; Renwick et al., 2000). In contrast, transient coupled AOGCM runs suggest that over the next 100 years or more, the mean strength of the westerlies actually will increase in the New Zealand region, particularly in winter (Whetton et al., 1996a; Russell and Rind, 1999; Mpelasoka et al., 2001). As a result, downscaled transient model results predict that rainfalls will increase in the west of New Zealand but decrease in the east (Mullan et al., 2001) as shown in Figure 12-2. Slightly larger temperature increases are predicted in the north of the country than in the south. Projected temporal changes in the westerlies before and after stabilization of climate also are likely to influence rainfall in southern Australia, especially in Tasmania.GCM simulations extending beyond the stabilization of GHG concentrations indicate that global warming continues for centuries after stabilization of concentrations (Wigley, 1995; Whetton et al., 1998; TAR WGI Chapters 9 and 11) but at a much reduced rate as the oceans gradually catch up with the stabilized radiative forcing. Importantly for the Australasian region, simulated patterns of warming and rainfall changes in the southern hemisphere change dramatically after stabilization of GHG concentrations. This is because of a reversal of the lag in warming of the Southern Ocean relative to the rest of the globe. This lag increases up to stabilization but decreases after stabilization (Whetton et al., 1998), with consequent reversal of changes in the north-south temperature gradient, the mid-latitude westerlies, and associated rainfall patterns.
Other reports in this collection