3.8 Changes in Extreme Events

3.8.1 Background

There is increasing concern that extreme events may be changing in frequency and intensity as a result of human influences on climate. Climate change may be perceived most through the impacts of extremes, although these are to a large degree dependent on the system under consideration, including its vulnerability, resiliency and capacity for adaptation and mitigation; see the Working Group II contribution to the IPCC Fourth Assessment Report. Improvements in technology mean that people hear about extremes in most parts of the world within a few hours of their occurrence. Pictures shot by camcorders on the news may foster a belief that weather-related extremes are increasing in frequency, whether they are or not. An extreme weather event becomes a disaster when society and/or ecosystems are unable to cope with it effectively. Growing human vulnerability (due to growing numbers of people living in exposed and marginal areas or due to the development of more high-value property in high-risk zones) is increasing the risk, while human endeavours (such as by local governments) try to mitigate possible effects.

The assessment of extremes in this section is based on long-term observational series of weather elements. As in the TAR, extremes refer to rare events based on a statistical model of particular weather elements, and changes in extremes may relate to changes in the mean and variance in complicated ways. Changes in extremes are assessed at a range of temporal and spatial scales, for example, from extremely warm years globally to peak rainfall intensities locally, and examples are given in Box 3.6. To span this entire range, data are required at a daily (or shorter) time scale. However, the availability of observational data restricts the types of extremes that can be analysed. The rarer the event, the more difficult it is to identify long-term changes, simply because there are fewer cases to evaluate (Frei and Schär, 2001; Klein Tank and Können, 2003). Identification of changes in extremes is also dependent on the analysis technique employed (Zhang et al., 2004a; Trömel and Schönwiese, 2005). To avoid excessive statistical limitations, trend analyses of extremes have traditionally focused on standard and robust statistics that describe moderately extreme events. In percentile terms, these are events occurring between 1 and 10% of the time at a particular location in a particular reference period (generally 1961 to 1990). Unless stated otherwise, this section focuses on changes in these extremes.

Global studies of daily temperature and precipitation extremes over land (e.g., Frich et al., 2002; see also the TAR) suffer from both a scarcity of data and regions with missing data. The main reason is that in various parts of the globe there is a lack of homogeneous observational records with daily | resolution covering multiple decades that are part of integrated digitised data sets (GCOS, 2003). In addition, existing records are often inhomogeneous; for instance as a result of changes in observing practices or UHI effects (DeGaetano and Allen, 2002; Vincent et al., 2002; Wijngaard et al., 2003). This affects, in particular, the understanding of extremes, because changes in extremes are often more sensitive to inhomogeneous climate monitoring practices than changes in the mean (see Appendix 3.B.2 and 3.B.4). Consistent observing is also a problem when assessing long-term changes in the frequency and severity of tropical and extratropical storms. Similar difficulties are encountered when trying to find worldwide observational evidence for changes in severe local weather events like tornadoes, hail, thunderstorms and dust storms. Analyses of trends in extremes are also sensitive to the analysis period, for example, the inclusion of the exceptionally hot European summer of 2003 may have a marked influence on results if the period is short.

Since the TAR, the situation with observational data sets has improved, although efforts to update and exchange data must be continued (e.g., GCOS, 2004). Results are now available from newly established regional- and continental-scale daily data sets; from denser networks, from temporally more extended high-quality time series and from many existing national data archives, which have been expanded to cover longer time periods. Moreover, the systematic use and exchange of time series of standard indices of extremes, with common definitions, provides an unprecedented global picture of changes in daily temperature and precipitation extremes (Alexander et al., 2006, updating the results of Frich et al., 2002 presented in the TAR).

As an alternative, but not independent data source, reanalyses can also be analysed for changes in extremes (see Appendix 3.B.5.4). Although spatially and temporally complete, under-representation of certain types of extremes (Kharin and Zwiers, 2000) and spurious trends in the reanalyses (especially in the tropics and in the SH) remain problematic, in particular before the start of the modern satellite era in 1979 (Marshall, 2002, 2003; Sturaro, 2003; Sterl, 2004; Trenberth et al., 2005a). For instance, Bengtsson et al. (2004) found that analysed global kinetic energy rose by almost 5% in 1979 as a direct consequence of the inclusion of improved satellite information over the oceans, which is expected to significantly affect analysed storm activity over the southern oceans, where ship data are sparse.

In this section, observational evidence for changes in extremes is assessed for temperature, precipitation, tropical and extratropical cyclones and severe local weather events. Most studies of extremes consider the period since about 1950 with even greater emphasis on the last few decades (since 1979), although longer data sets exist for a few regions, enabling more recent events to be placed in a longer context. The section discusses mostly the changes observed in the daily weather elements, where most progress has been made since the TAR. Droughts (although they are considered extremes) are covered in Section 3.3.4 as they are more related to longer periods of anomalous climate.