Summary: Climate modeling and climate change scenarios for the region are relatively well developed. For Australia, the recently revised (1996) CSIRO scenarios for 2030 indicate temperature increases of 0.3-1.4�C and rainfall changes of up to 10% in magnitude (decreases in winter, increases or decreases in summer, and overall a tendency for decreases). The projected changes for 2070 are about twice those of the 2030 changes. Increases in the intensity of heavy rainfall events are indicated. For New Zealand, the temperature increases are expected to be similar to those for Australia, but the recent revision indicates the possibility of an increase in westerly winds (unlike the decreases of previous scenarios)-and hence precipitation increases in the west and precipitation decreases in the east. The changes in scenarios serve to caution against overinterpretation of impact studies based on any single scenario.
The IPCC has concluded that climate models at present provide useful predictions at the global and continental scale, but as yet allow little confidence at subcontinental scales (IPCC 1996, WG I, Section 6.6.3; Annex B of this report). A considerable amount of global climate model development, testing, and intercomparison has been done in the Australasia region (Garratt et al., 1996; Gordon et al., 1996; Whetton et al., 1996b; Whetton et al., 1996d). This research currently provides the best source of climate change scenarios for the region. Two factors that are important in the region's temperature changes are the moderating role of the extensive Southern Hemisphere oceans and the relative absence of aerosols that partially offset greenhouse gas warming in the northern hemisphere (Ayers and Boers, 1996; IPCC 1996, WG I, Section 184.108.40.206; Rintoul et al., 1996).
To facilitate impact and policy studies, a series of scenarios for the region have been progressively developed in line with the development of climate modeling research. Scenarios for Australia were released in 1987 (Pittock, 1988), in 1990 (Whetton and Pittock, 1990), in 1991 (CSIRO, 1991; Whetton et al., 1992), in 1992 (CSIRO, 1992; Mitchell et al., 1994; Whetton et al., 1996a), and in 1996 (CSIRO, 1996a). Similar scenarios have been released for New Zealand (Salinger et al., 1987; Salinger and Hicks, 1990a, 1990b; Salinger and Pittock, 1991; Mullan, 1994-also described in Whetton et al., 1996a). The integrated impact assessment models being developed for New Zealand ('CLIMPACTS,' Kenny et al., 1995) and Australia ('OzClim,' CSIRO, 1996b) both include a scenario generator to enable the potential impacts of multiple scenarios to be investigated.
The majority of recently published regional impact studies are based on the scenarios of CSIRO (1992) for Australia and Mullan (1994) for New Zealand. These scenarios used spatial patterns of regional climate change derived from five general circulation model (GCM) experiments-namely, J1, F1, A3, H3 of Table 1-1, and the Bureau of Meteorology Research Center (BMRC) experiment of Coleman et al. (1994). For 2030, these gave warmings in Australia of 0.5-2.5�C for inland areas, 0-1.5�C for northern coastal areas, and 0.5-2.0�C for southern coastal areas, and warmings in New Zealand of 0.5-2.0�C in inland Canterbury and Otago and 0.5-1.5�C elsewhere. The Australian summer precipitation projections ranged from little change to as much as a 20% increase, and winter precipitation projections showed as much as 10% change-but the direction of change depended on the district. In New Zealand, regardless of season, precipitation increases from zero to about 20% were projected-except around Wellington, the east coast of the North Island, South Canterbury, and Otago. The projected changes for 2070 were about twice the magnitude of the 2030 changes.
The CSIRO (1992) and Mullan (1994) scenarios took into account the IPCC (1990) range of plausible greenhouse-gas emission scenarios and sensitivity of the global climate to increased greenhouse forcing, but the GCMs used were of a "slab ocean" type-that is, they did not include an interactive deep ocean. Very recently, the application of coupled ocean-atmosphere GCMs and the use of updated IPCC (1996a) global warming scenarios has led to significant shifts in the projected changes of climate (CSIRO, 1996a). The CSIRO (1992) and Mullan (1994) scenarios used the same five slab-ocean GCMs (except for an improved version of F1), and five coupled-ocean GCMs (X10, X2, X8, and X6 of Table 1-1, and another MPI experiment as in Lunkeit et al., 1994). The marked differences in scenarios from the coupled models in the Australasian region highlight the need for caution in the use of climate scenario information for impact assessment.
In the new CSIRO (1996a) scenarios, the projections of warming for Australia have been reduced, mainly due to the downward revision of global warming estimates. The new scenarios for 2030 are 0.4-1.4�C for inland areas, 0.3-1.0�C for northern coastal areas, and 0.3-1.3�C for southern coastal areas. The use of coupled ocean-atmosphere models in addition to the older non-coupled "slab ocean" models has led to major changes in summer rainfall scenarios, which for 2030 now vary from -10% to +10%, unlike the increases projected by the slab models alone. This difference appears to be due to the coupled models' much slower rate of warming in the higher latitudes of the Southern Hemisphere (Whetton et al., 1996b). Winter rainfall change scenarios for 2030 are similar in coupled and slab models and range from decreases of up to about 10% over inland areas to increases of a similar magnitude in the far south (mainly Tasmania). Related scenarios based on the results of a regional model (DARLAM) nested in the CSIRO slab-ocean GCM (experiment F1, Table 1-1) also have been developed (Hennessy et al., 1997; Whetton et al., 1997).
It should be noted that the CSIRO (1996a) scenarios continue to use results from the older slab-ocean climate models as well as from the newer coupled models. This is because there are still some doubts about the reliability of the scenarios from the coupled models (Whetton et al., 1996b). These doubts partly relate to the observed latitudinal gradient of warming in the Southern Hemisphere (IPCC 1996, WG I, Section 220.127.116.11), which is more like that in the slab-ocean GCMs, and to the simulated rainfall decreases in summer over Australia-which are contrary to the observed small increases this century. This disagreement could be because the observed trends are not predominantly greenhouse-related and may be influenced by natural multidecadal fluctuations in the coupled ocean-atmosphere system.
Alternative scenarios for New Zealand based on the newer coupled model results are in preparation. Significant differences in the projected patterns of rainfall change are expected because the coupled models simulate an increase in the strength of the mid-latitude westerly winds, which is the reverse of the previous slab-model based scenarios. Stronger westerlies are likely to show a greater tendency for rainfall increases on the wet western side of the Southern Alps and decreases on the dry eastern plains.
Any overall increase in temperature will tend to cause an increase in the frequency of extremely high temperatures and a decrease in the frequency of extremely low temperatures (e.g., frost) (IPCC 1996, WG I, Section 6.5.7). A similar effect will apply to areas of overall rainfall increase, and vice versa for areas of overall rainfall decrease. In addition, global climate models tend to predict that in a warmer climate there will be more high-intensity rainfall events, with more severe droughts and floods in some places but less severe droughts and floods in other places (Fowler and Hennessy, 1995; IPCC 1996, WG I, Section 6.5.7). Where interannual variability in rainfall is high, there may be more sensitivity to changes in the frequency of extreme events than to small changes in the mean climate (IPCC 1996, WG II, Section 2.7.1).
Soil moisture will be directly affected by rainfall changes, including the replenishment of groundwater by more frequent high-rainfall events. Moreover, higher temperatures will increase the potential for evapotranspiration, causing faster drying of the soil.
Of particular concern for the Australasian region are possible changes to the region's major weather systems-especially the timing, intensities, and locations of the tropical monsoonal and cyclone systems; the locations and intensities of mid-latitude weather systems and the subtropical anticyclone belt; and the frequency, intensity, and bias of the El Ni�o-Southern Oscillation. Unfortunately, climate models do not at this stage provide firm projections for any of these key features (e.g., Pittock et al., 1995; IPCC 1996, WG I, Sections 6.4.4, 6.5.4; Katzfey and McInnes, 1996; Pittock et al., 1996; Suppiah et al., 1996), although there are some tentative indications that tropical cyclones could increase their maximum potential intensities by as much as 10-20% under doubled CO2 conditions (Henderson-Sellers and Zhang, 1997; Holland, 1997). The use of nested regional models may provide scenarios of tropical cyclone changes in the future (Suppiah et al., 1996; Walsh and Watterson, 1997).
There are large differences between the results of the different global models, especially at the subcontinental scale (IPCC 1996, WG I, Section 6.6; Whetton et al., 1996d), and any regional circulation changes could impose large changes in temperature and rainfall on top of the continental-scale changes indicated above.
Other reports in this collection