While the changes in both the mean and higher order statistical moments (e.g., variance) of time-series of climate variables affect the frequency of relatively simple extremes (e.g., extreme high daily or monthly temperatures, damaging winds), changes in the frequency of more complex extremes are based on changes in the occurrence of complex atmospheric phenomena (e.g., hurricanes, tornadoes, ice storms). Given the sensitivity of many exposure units to the frequency of extreme climatic events (see Chapter 3 of TAR WG II, Table 3.10 (Carter and La Rovere, 2001)), it would be desirable to incorporate into climate scenarios the frequency and intensity of some composite atmospheric phenomena associated with impacts-relevant extremes.
More complex extremes are difficult to incorporate into scenarios for the following reasons: (1) high uncertainty on how they may change (e.g., tropical cyclones); (2) the extremes may not be represented directly in climate models (e.g., ice storms); and (3) straightforward techniques of how to incorporate changes at a particular location have not been developed (e.g., tropical cyclone intensity at Cairns, Australia).
The ability of climate models to adequately represent extremes partially depends on their spatial resolution (Skelly and Henderson-Sellers, 1996; Osborn, 1997; Mearns, 1999). This is particularly true for complex atmospheric phenomena such as hurricanes (see Chapter 10, Box 10.2). There is some very limited information on possible changes in the frequency and intensity of tropical cyclones (Bengtsson et al., 1996; Henderson-Sellers et al., 1998; Krishnamurti et al., 1998; Knutson and Tuleya, 1999; Walsh and Ryan, 2000); and of mid-latitude cyclones (Schubert et al., 1998), but these studies are far from definitive (see Chapter 9, Section 9.3.6, and Chapter 10 for discussion on changes of extremes with changes in climate).
In the case of extremes that are not represented at all in climate models, secondary variables may sometimes be used to derive them. For example, freezing rain, which results in ice storms, is not represented in climate models, but frequencies of daily minimum temperatures on wet days might serve as useful surrogate variables (Konrad, 1998).
An example of an attempt to incorporate such complex changes into climate scenarios is the study of McInnes et al. (2000), who developed an empirical/dynamical model that gives return period versus height for tropical cyclone-related storm surges for Cairns on the north Australian coast. To determine changes in the characteristics of cyclone intensity, they prepared a climatology of tropical cyclones based on data drawn from a much larger area than Cairns locally. They incorporated the effect of climate change by modifying the parameters of the Gumbel distribution of cyclone intensity based on increases in tropical cyclone intensity derived from climate model results over a broad region characteristic of the location in question. Estimates of sea level rise also contributed to the modelled changes in surge height. Other new techniques for incorporating such complex changes into quantitative climate scenarios are yet to be developed.
Other reports in this collection