THE PURPOSE of this section is to examine whether any trends can be recognized in disasters that have involved coasts and population over recent decades.

There are several databases on disasters, usually focusing on specific themes. A brief introduction to the subject is given by IFRCS, 1996. For the current study, we adopted the database assembled by the Office of the US Foreign Disaster Assistance, among others because all types of disasters are covered. Information is more complete on the events since 1964 (OFDA, 1996), and the records exclude disasters within the US and its territories. Basically it includes only "major" disasters, based on losses and number of people affected, homeless or killed. The criteria are less strict for small island countries, with the consequence that they tend to be over-represented.

Table 4 below quantifies the loss of life associated with disasters this century. Disasters have been grouped in categories as follows:

• "Sea" (555 disasters), caused by extreme factors associated with the seas and oceans, mainly tropical cyclones, hurricanes, tsunamis and typhoons)
• "Land" (1918 disasters) including avalanches, cold waves, drought, earthquakes, fires, floods, heat waves and landslides)
• "Population": (1043 disasters) mostly events due to war and diseases, including internally displaced persons and refugees, epidemics, famine and food shortage
• "Others": (322 disasters) covering power shortages, storms, volcanic eruptions [7] and unusual phenomena.

It appears that man-made "population" disasters tend to create more victims than both sea and land-bound disasters. As an example, wars and political disputes have, over the recent decades, replaced drought as one of the main causes of food shortages. Again, note the marked positive skew of the distribution of all types of disasters.

The years since 1964 [8] have witnessed an upward trend in population and land-bound disasters. Sea-bound disasters have shown a positive, but much less marked, trend as well, with a drop around 1990. Figure 6 shows the relative importance of the four categories: as noted, population-related disasters have been on the increase , while coastal areas have been losing importance as disaster areas. This is difficult to reconcile with the hypothesis that growing populations along the coasts would have become more vulnerable than in the past, particularly if one considers that the 124 million people living in the 15 main cities in 1965 (New York, Tokyo, Shanghai, London, Paris, Beijing, Buenos Aires, Los Angeles, Osaka, Mexico, Moscow, Essen, Chicago, Sao Paulo, Calcutta) constituted a significantly less "coastal" and vulnerable "environment" target for extreme events than the current 220 million persons living in the mostly coastal megacities. Furthermore, cities evolve over time, and in many cases could be expected to adjust to gradual change in sea-level over the next century - of course up to certain limits impossible to specify.

If methodological problems and sampling problems can indeed be excluded (see, for instance Pielke and Landsea, 1997), and provided epidemics, famine, etc. are not concentrated in the main world urban centres, one is bound to assume that cities, even coastal ones, constitute a rather safe environment.

Figure 6. Annual number of disasters by category
Based on data in OFDA, 1996

Figure 7. Total annual damage in 1992 constant US$ due to the "land" and "sea" based disasters, together with their linear trends
Based on data from OFDA (1996) and the implicit price deflator (Umich, 1997)

In Figure 6, an attempt was made to extrapolate the total annual damage due to "land" and "sea" disasters. Extrapolations of this kind have a large statistical variation, but they illustrate the fact that it is mainly the increase in wealth which leads to increased damage, rather than the increased frequency of disasters. In addition, there is no indication as to whether the "wealth structure" (wealth is now mainly concentrated in buildings and infra-structure, and to some extent in land) will not undergo significant qualitative changes (for instance, wealth could be in electronic equipment, or databases, knowledge etc.).

Notes
7. Volcanoes tend to be along the coasts, and many small islands are actually of volcanic origin, but there is no absolute association.

8. The trends exist before 1964, but only the period after 1964 is covered because observations are more complete beginning around that year.