There has been a huge increase in economic and insured losses from earthquakes. Table 1 clearly shows the rising trend to ever more and ever greater losses - a trend which is true of all natural catastrophes. Mankind has encroached ever further into exposed areas. An increasing number of people and values are concentrated in a few highly exposed cities. An earthquake that even in the 1960s would have affected hardly anyone may lead to a catastrophe today.
The maximum loss potentials have mushroomed to almost astronomical heights as a result of this development. For example, a large earthquake in Tokyo today could result in economic losses of over US$1,000 billion.
The insurance industry is disproportionately affected by this increase. The reason for this is to be found primarily in rising insurance densities, a process that will be further accelerated in the future as a result of the deregulation of insurance markets and governments shifting the onus of payments to the private sector. For example, Belgium is considering the possibility of offering widespread earthquake covers for residential buildings.
This trend of increasing quake risks continued in 1999. With the earthquakes in Armenia/Colombia, Huajuapan/Mexico, Izmit and Düzce/Turkey, Athens/Greece and Chi-Chi/Taiwan, 1999 was the worst affected year in terms of earthquake catastrophes since 1976, notwithstanding the individual events at Northridge/California in 1994 and Kobe/Japan 1995.
Findings from 1999 earthquakes
The earthquakes in 1999 should be used to help us be better prepared for such events in the future. In particular, the earthquakes in the second half of the year were perfect examples of certain aspects of earthquake risk today, such as predictability, the consequences of the enormous growth of cities and the susceptibility to losses of high technology.
Although it was accepted that the area around Izmit was a “seismic gap” and an earthquake of this magnitude was foreseeable, no practical consequences were drawn from this well-founded scientific knowledge, even though the area affected accounts for 40% of the country's industrial production. The existence of another seismic gap to the south of the 10-million city of Istanbul sends a clear warning to all parties concerned of the need to invest in loss prevention if we are to avoid an even more disastrous repeat of events at Izmit. The situation, in Athens and Taiwan were quite different. Both events took experts by surprise and should serve as a warning that even faults assigned a low or uncertain level of activity can “spring a nasty shock”.
The massive growth of cities not only contributes to the enormous loss potentials in terms of the resulting concentrations of values but also has a negative effect on the loss susceptibility of urban areas. If Athens had not grown to its present size, the earthquake of 7 September would have only been an “episode on the fringes”. As it is, however, the economic losses of US$3.7 billion make it a supreme example of the loss potential of even moderate earthquakes. The problems associated with – in some cases uncontrolled – growth were demonstrated even more clearly (in line with the size of the city affected) in the case of the Izmit earthquake. The necessity of creating housing for millions of “migrants” to the Istanbul area in the space of just a few years resulted in an almost inevitable loss in the quality of construction. The construction boom and the criminal energy of certain building companies led to a situation where all regulations governing responsible construction were cast aside. Unfortunately, the lessons of the Izmit earthquake can be seen everywhere.
Modern high technology has resulted in completely new loss potentials, as the high losses in Taiwan's semiconductor industry amply demonstrated. Although the Science Park at Hsin-chu is more than 100 km from the earthquake fault and suffered only minor material damage, insurers had to pay out several hundred million US dollars. The main reason for this huge amount was the breakdown in the power supply and the resulting business interruption and the loss of chip production.
The path to greater earthquake safety and to a substantial reduction in earthquake losses, as defined on the basis of the 1999 earthquake year, can be summarised in four points.
Assessing exposure in weak
In addition to the findings that can be gained from today's earthquake catastrophes, seismo-geological research at universities provides an essential contribution to better assessment of global loss potentials.
The reinsurance industry today is using detailed risk models to better assess the earthquake risk. The essential element of all models is as exact assessment as possible of earthquake activity as well as the frequency and strength of the earthquake that can be expected. In areas of high seismicity, records on historical events and instrumental data exist and usually a lot of time and money are invested in researching active structures as a result of this high exposure. In areas of low seismicity on the other hand, only a limited amount of information is usually available. However, it is precisely these areas where buildings, infrastructures and risk management are not designed for earthquakes, thus making a failure of these aspects almost inevitable. The loss potentials here are disproportionately high compared with earthquakes in other areas. An example of an area with low seismicity and enormous loss potentials is central Europe. The lack of knowledge about occurrence probabilities and magnitudes of large earthquakes in central Europe leads to an enormous level of uncertainty with regard to exposure estimates. The loss potentials, estimated in Allmann et al (1998) for Cologne, perfectly illustrate the problems involved (table 2).
The starting point for this scenario analysis was studies from the paleoseismological work of Camelbeek & Meghraoui (1996) and Ahorner (1996). The hypothetical earthquakes represent possible scenarios on the basis of our incomplete knowledge at this time. However, statements on activity rates and maximum strength of the earthquakes would be very speculative without further research.
Rather like in, for example, North America east of the Rocky Mountains, in Australia or in Greater Basle, the earthquake risk in Germany is a typical problem of low occurrence probability coupled with potentially devastating effects. In all the aforementioned regions considerable funds have been, and continue to be, invested in special research projects. In Germany, such projects are just getting off the ground.
Even if one disregards the extreme scenarios outlined above, research projects with the objective of achieving a better limitation of the exposure situation appear essential. This research can be summarised as follows:
The estimates for Cologne can, to a lesser extent, also be applied to the Frankfurt and Stuttgart areas.
In 1999, Belgium also became a focal point of interest for the insurance industry. The market pressure in Belgium to offer earthquake insurance on a broad basis for residential buildings has led to increased interest in the exposure situation there. Like Cologne, Belgium is characterised by a weak seismicity. Although there is a general estimate of exposure, occurrence probabilities of large events and their possible maximum magnitudes are extremely uncertain. For the insurance industry, this can mean differences of billions of dollars in the loss potentials. This not only affects the price level of insurance but also influences the possibility of providing an insurance solution in general. The economic importance of research and the improved level of knowledge it entails cannot be overstated.
Another example to illustrate the loss potentials of earthquakes in low seismicity regions is a scenario based on the repetition of the Neulengbach (1590) earthquake close to Vienna in Austria. The Isoseimal map of this event by Hammerl and Lenhardt (1997) was used to define the area affected by that event and to evaluate the expected intensity (MMI) for each postcode in Austria. Assuming an amount of 1.66 million Austrian Shilling for each resident in the respective postcodes to represent the value of the housing stock we derived the residential values in each intensity zone (table 3). By using the same loss ratios as described above for the Cologne scenario, an expected loss of 140 billion Austrian Shilling (US$10 billion) has been calculated. Although this result is only a rough estimate, it highlights the possible loss potentials in regions of low to moderate seismicity. By using an historic scenario the question is avoided, whether the worst case scenario is only an academic speculation or can really happen. But even for the region where the Neulengbach event occurred 400 years ago, only little is known about the potential of devastating earthquakes.
That events which are theoretically possible, but difficult to assess without the aid of further research, can really happen was shown to be true in 1999 with the events at Athens/Greece and Chi-Chi/Taiwan, Although both faults were already known, the lack of micro-seismicity and more detailed investigations meant that a significant earthquake was not expected. If Athens and Taiwan were not exposed to other more active faults, the consequences of a complete lack of adequate building designs would have been much more serious than they actually were (such as in Tangshan/China 1976).
Last year was another seriously affected by earthquake catastrophes, thus continuing the trend over the last few decades towards increasing economic and insured losses. The loss potentials of major earthquake catastrophes today have reached almost astronomical heights. The reduction of loss potentials coupled with improved loss potential assessment is therefore becoming increasingly important. Catastrophes such as those in 1999 should be used more than ever as something to learn from. However, new findings also have to be implemented in practice.
In many regions of low seismicity, especially in central Europe, knowledge of earthquake exposure is nowhere near adequate. Realistic loss potentials can only be estimated with an enormous degree of uncertainty and in some cases such estimates do not amount to more than mere speculation. In view of the high loss potentials and their economic importance, an increase in research activity in these regions appears absolutely essential.
Alexander Allmann and Anselm Smolka are seismologists in Munich Reinsurance Company's geoscience research group.
E-mail: Aallmann@munichre.com; Asmolka@munichre.com
Ahorner, L. 1998, Möglichkeiten und Grenzen paläoseismologischer Forschung in mitteleuropäischen Erdbebengebieten. In: Paläoseismologie, Eurocode 8 und Schwingungsisolierung, DGEB-Publikation Nr. 9, 9-44
Allmann A., Rauch E., Smolka A.; 1998 – New paleoseismological findings on major earthquakes in Central Europe: possible consequences for the earthquake potential in Germany. Proceedings 11th European Conf. Earthqu. Eng. Rotterdam: Balkema
Camelbeeck T, Meghraoui, M 1998. Geological and geophysical evidence for large paleo-earthquakes with surface faulting in the Roer Graben. Geoph.J.Int. 132: 347-362
Hammerl C., Lenhardt W., 1997. Erdbeben in Österreich, Leykam Verlag, 103
The July 2000 EuroConference on global change and catastrophe risk management: earthquake risks in Europe was organised by the International Institute for Applied Systems Analysis and was sponsored by the European Commission within the Training and Mobility Researchers (TMR) Program.