Figure 8-2: Schematic showing effect on extreme temperatures when (a) mean temperature increases, (b) variance increases, and (c) when both mean and variance increase for a normal distribution of temperature (TAR WGI, Figure 2.32).
Figure 8-2: Schematic showing effect on extreme temperatures when (a) mean temperature increases, (b) variance increases, and (c) when both mean and variance increase for a normal distribution of temperature (TAR WGI, Figure 2.32). |
Weather-related events of all magnitudes resulted in US$707 billion in insured and uninsured economic losses between 1985 and 1999 (Munich Re, 2000). A longer term comparison of large catastrophic events over the past 50 years reveals that economic losses (adjusted for inflation) increased by a factor of 10.3 (Figure 8-1). Over this same period, population grew by a factor of 2.4.
One of the vexing dilemmas in analyzing such historical data is disentangling causal factors related to human-induced climatic change, natural variability, and those having to do with human activity that could accelerate or dampen measured impacts. Numerous human factors are in operation that contribute to the upward trends in real economic losses, including population growth, rising standard of living, urbanization and industrialization in high-risk regions, vulnerability of modern societies and technologies, environmental degradation, penetration of insurance, and changing societal attitudes toward compensation (the latter two factors may lead to an increase in losses reported). Data on the numbers of events also show an increase in many cases. The number of disasters (defined as annual requests from states for federal disaster declarations) has roughly doubled in the United States since the early 1980s (Anderson, 2000). It is relevant to note here that such requests involve considerations of significant social effects (Kunkel et al., 1999); as a consequence, it is an indirect and subjective proxy for the frequency of events.
Growth trends in non-climate-related losses have been relatively constant over the past 3 decades. Losses from human-induced catastrophes have remained relatively constant (Swiss Re, 1999a). Earthquake losses have increased, but more slowly than weather-related losses (Bruce et al., 1999). The number of disasters causing more than 1% GDP damage to affected countries has increased two to three times as rapidly for weather-related disasters as for earthquakes in the period 1963-1992 (United Nations, 1994).
Insurers have pointed out that local environmental factors such as soil degradation, loss of biodiversity, lack of drinkable water, pollution, deforestation, forest degradation, and land-use changes can amplify the impacts of weather-related catastrophes (Zeng and Kelly, 1997). As an illustration, the extent of flood losses from Hurricane Mitch was attributed in part to deforestation in Central America.
Attempts to analyze the underlying causes of trends in natural disasters also must allow for the effects of human activities that offset growth factors (Kunkel et al., 1999). A considerable leveling off or reduction in loss of life during U.S. disasters is one indicator that mitigation has been effective (Easterling et al., 2000a). Loss-reduction efforts—typically unaccounted for in analyses we have seen—include considerable efforts to avert or reduce natural disaster impacts (e.g., coastal protection structures along coastlines; cloud seeding to deflect hailstorms; improved building codes; tightened land-use zoning; enhanced fire-suppression capacity; improved weather forecasts and early-warning systems; and improved disaster preparedness, response, and recovery). Within the insurance arena, increasing deductibles (the initial tier of loss costs paid for by the insured) and withdrawal of coverage from particularly high-risk areas have reduced observed losses. The literature has not attempted to quantify the contribution of these activities.
The relative contributions of human and climatic factors to the changing patterns of losses varies, depending on place and type of event (see Table 8-1; also see Easterling et al., 2000a,b for a review). U.S. studies have found that demography largely explains increases in losses for hurricanes, wind, hail, and tornado events, whereas winter storm damage has mixed causation (Pielke and Landsea, 1998; Changnon, 1999; Changnon and Changnon, 1999; Kunkel et al., 1999). In addition, decadal-scale trends have been discerned for tropical cyclones. There is good evidence that the intensity and frequency of precipitation and flood-related extreme events in the United States is increasing (Zeng and Kelly, 1997; Karl and Knight, 1998; Pielke and Downton, 2000). This trend also has been found for precipitation in many other parts of the world (see Chapter 3). In a study of hailstorms in France, Dessens (1995) used insurance loss information as a proxy for storm occurrence and found a statistically significant upward trend between 1946 and 1990.
In one global analysis, Munich Re (1999b) estimates that economic losses from large natural disasters increased two-fold between the 1970s and 1990s, after correcting for inflation, insurance penetration and pricing effects, and increases in the material standard of living. A similar result was reported for UK buildings in the SAR (Dlugolecki et al., 1996).
Based on the findings of TAR WGI, the information summarized in Table 8-1, and the analysis presented above, we conclude that some part of the upward trend in the cost of weather-related disasters illustrated in Figure 8-1 is linked to socioeconomic factors (increased wealth, shifts of population to the coasts, etc.) and some part is linked to climatic factors such as observed changes in precipitation and drought events. There are regional differences in the balance of these two causes.
|
Other reports in this collection |
|
