Because of nonlinear relationships, an increase in variability can result in a substantial increase in the frequency of extreme impacts. If a climate element exceeds an acceptable risk threshold (e.g., when the design risk threshold for water storage is exceeded and water shortage is experienced with higher frequency), vulnerability will become "unacceptable." One issue in adaptation is the level at which to set acceptable risks in the future. Stakeholder-determined thresholds are an emerging area of research in Australia (see Section 12.1), and methods to evaluate stakeholder and institutional learning in response to changing climatic hazards are being developed (see Bakker et al., 1999; see also <http://www.eci.ox.ac.uk>). Decision analytical techniques are described below (Section 2.7). An alternative is an inverse approach that focuses on sensitivity to present risks, characterization of the kinds of changes in hazards that would have large effects, and evaluation of response capacities (Downing et al., 1999a).
Research on discrete climatic events is an area that also needs further research. Present GCM resolutions have not achieved the ability to estimate the intensity, route, and frequency of discrete events such as hurricanes (or tropical cyclones) (TAR WGI Sections 9.3.6 and 184.108.40.206). Though there are some indications from GCMs that ENSO-like conditions will become more persistent with global warming (Timmermann et al., 1999), it is still difficult to incorporate these estimates into vulnerability assessments (TAR WGI Sections 220.127.116.11 and 18.104.22.168).
Empirical/analog methods are suitable for assessment of discrete events. Such methods were applied for detailed analyses of damages incurred by ENSO in 1997-1998, as well as the 1998 cyclones in Bangladesh. This method is applied to "if-then" (i.e., if climate change occurs, then such and such impacts may be induced) simulations. For example, analogs from the 1930s Dust Bowl period detailing water shortages and reductions in agriculture yields have been used to simulate the impacts of climate change in the U.S. corn belt (Rosenberg, 1993).
Because unique or singular events, referred to as fiasco scenarios (see Section 22.214.171.124)such as changes in the thermohaline circulation (Broecker, 1997) and potential destabilization of the West Antarctic Ice Sheet (Oppenheimer, 1998)have not been proven implausible, there is a need for further studies of potential catastrophic events and unacceptable impacts. However, limited knowledge of such large-scale impacts poses a challenge; to date, systematic vulnerability assessments have not been carried out.
Other reports in this collection