Prediction and Perception of Natural Hazards

In the streets of Florence in the thirteenth century, though prohibited by law, dice players used to shout “zara” when the total value of the three dices rolling over the table showed the lowest probability sets of results (less than or equal to seven, or greater than or equal to fourteen, meaning “null try”): “quia cum tribus taxillis raro veniunt sex vel quinque et rarius quator vel tria...eodem modo accidit inter quator decim et decern octo” (2). From the arabic “azzahr”, the dice, the sense of rare events, not involving any special connotation of danger or risk, came into French as “hasard”, into Italian as “azzardo” and into English as “hazard”.

There are many factors that can influence the seriousness of a disaster, ranging from the degree of preparedness and planning to the availability of relief and post-disaster assistance. In the case of tropical storms (hurricanes, typhoons and cyclones), however, a primary agency of the disaster is the failure and unservicability of buildings and structures. Amongst these are homes, refuges, hospitals, schools, industrial buildings, communication towers, and power lines. Apart from the personal injuries which the structural damage causes, it deprives people of shelter, disrupts essential post-disaster services such as hospitals and communications, cripples important contributors to the economy, such as the manufacturing, tourist and agricultural industries and impedes relief and recovery. The evidence suggests that most of this structural damage is preventable at little or no cost.

When reflecting on appropriate IDNDR-strategies, we must seek relief programs that avoid dependency and inactivity. To put it positively, we must seek programs that support indigenous efforts to prevent disasters and to overcome hazards without continuous outside support.

This paper focuses on natural hazard prediction and associated responses in the context of large scale modern urbanization. It reviews evidence from a selection of large-city earthquake, storm and wildfire disasters and focuses on implications of the Loma Prieta earthquake of October 17, 1989. It is suggested that, in large-city disasters of the future, issues of prediction and response are likely to be over-shadowed by other pressing management concerns.

Generally scientists communicate the results of their researches to restricted environments of specialists, well aware of the specialized scientific languages and informed of the limits of the uncertainties implied by the results of such research works. However in some cases scientists must transmit the conclusions of their researches to larger groups of recipients, including people not familiar with the language of science. In such cases, given the fact that the research data utilised are affected by errors and hence are characterised by certain probability values, non-trivial problems of communication could arise.

There is a current opinion that modern civilization has reduced the risk of natural hazards. A thorough analysis of empirical data gathered over the years, however, argues that our society continues to be highly vulnerable to natural disasters that are more destructive then ever. My estimate shows that average losses inflicted by these disasters in the former Soviet Union have now exceeded two percent of GDP per year. Therefore, economic and intellectual input into forecasting should be increased. More accurate and sophisticated forecasting would reduce the number of victims and their losses from natural disasters.

This overview paper addresses the issue of the uncertainties we face in the predictions -for purposes of natural disaster reduction — of atmospheric phenomena ranging from large scale extra-tropical storms and droughts to tropical cyclones (hurricanes and typhoons) and tornadoes. I will address each phenomena summarizing the prediction capabilities today and, where possible, give some idea of the outlook for the future.

Three studies (Hanson, Vitek, and Hanson, 1979; Glass et al., 1980; Hodler, 1982), and to a lesser degree a fourth (Taylor, Zurcher, and Key, 1970), provide most of the survey data documenting public response to tornado warnings.

After the model has been calibrated for a given area, hundreds of hypothetical hurricanes are “run through” the model to determine each storm surge inundation pattern. The hypothetical hurricanes are systematically chosen to reflect variations in intensity, point of landfall, angle of approach and forward speed. Each model run presents the storm surge height for a number of model grid points, both along the open coast and well inland.

Non structural actions for flood control have had a very consistent development in the last twenty years for many reasons and in particular:

According to Burton, Kates and White, (1978) the impact of natural hazards is a function of the benefits associated with living in a risk zone minus the costs of damage and of adjustment to the hazard. Alternatively, UNDRO (1982) argued that the sum total of risk is a product of the magnitude and extensiveness of the natural hazard impact, the number and size of elements at risk, and their vulnerability in terms of probable levels of damage and destruction.

In many countries landslides are widespread in space and time as to constitute the natural disaster which generates a yearly loss of property greater than that derived from any other catastrophe such as earthquakes, tornadoes and floods. This is largely true for Italy where the interaction of geological, geomorphological and climatic factors has led to a geomorphic evolution of the slopes mainly controlled by mass-movement. Hence, throughout the hilly and mountainous areas of the country, every year hundreds of failures take place damaging lifelines, dwellings and crop yields (Cotecchia, 1978; Canuti, 1982).

European rivers tend to have long and well documented history of floods. The Danube, Europe’s second largest river after the Volga, flows through or marks the border of eight states: Austria, Bulgaria, Czechoslovakia, Hungary, Romania, the Soviet Union, West Germany and Yugoslavia. As an international river, 2,857 km long, and with a catchment area of 817,000 km², it poses many ecological and political problems.

We are wrapping up a three year project assessing flash flood hazard in the United States. Our effort has three main components: 1) defining flash floods as opposed to slow rise floods; 2) reviewing the experiences of communities that are flash flood mitigation success stories; and 3) summarizing the advantages and disadvantages of warning systems. My presentation focuses on the third mission. Our work concentrates on flash floods in the United States but I am sure that some of our findings can be directly applied to international contexts and to landslide mitigation.

A great diversity of methodologies relative to landslide prediction have been developed in recent years and especially, hazard and risk mapping have now become usual techniques. According to Hartlen and Viberg (1988) the complete landslide hazard evaluation should provide answers at least to the following questions: where will the landslides occur? which will be its volume and how far will it travel? when will it happen? what kind of movement will take place? how fast will it move? Even though all the questions are relevant, only the first two questions will be discussed in the paper.

During the last decade there has been a considerable development and an intense activity of Earthquake Engineering all over the world. The general objective of that relatively new field of engineering is the study of the behaviour of buildings and other engineering structures (bridges: towers: reservoirs: etc.) under earthquake action. For the attainment of the above general objective, earthquake engineering follows two main directions: one towards the design of new earthquake resistant structures, and the other towards the protection and strengthening of existing structures. In following these two broad directions, earthquake engineering makes use of all available scientific knowledge that can be of help, not only from the related fields of structural and geotechnical engineering, but also from other fields like geology, geophysics and seismology. In this context, an interesting question arises whether prediction of earthquakes, a subject widely discussed at present, can be of use to earthquake engineering in its effort to achieve the above objectives.

Efforts in the United States continue to communicate information about earthquake risk to the public. These efforts fall into two categories. First, risk is communicated to inform and educate citizens about general earthquake risk. Second, risk is communicated in reference to the prediction of warning of a specific impending earthquake. Central to both risk communication types are the perceptions which the public forms about earthquake risk since these perceptions direct public behavior.

The engineering problem of understanding the collapse of a building during a major earthquake is greatly eased if the actual seismic motion on the site is known. Whenever a sufficient number of strong motion records is available, as in the case of the november 23, 1980 Irpina (Italy) earthquake, we can realistically simulate it for frequencies up to 1 Hz using “complete” synthetic seismograms. In our case we apply a normal mode summation technique. Through a simple minimization process we obtain a source model that reproduces the observed records. Then we use it to estimate the seismic ground motion at other sites. We also investigate the stability and uncertainties in our solution.

Volcanic activity resumed at Mt. St. Helens on March 20, 1980, after 123 years of quiet. Over a period of five days, the threat of a major eruption increased and subsided. Media coverage of the volcano was initially intense, but gradually declined and nearly disappeared by the second week of May (Earle, Southwick and Lindell 1981). In addition to fluctuations in the severity of the threat, there were uncertainties about the areas at greatest risk. During the initial emergency period, the Lewis River Valley on the south side of the mountain was thought likely to be the area of most severe impacts. Later the formation of a bulge on the northwest side of the volcano diverted attention to the Toutle River Valley, which ultimately bore the brunt of the catastrophic explosion on May 18th. Since that time, steam, ash and effusive (dome building) eruptions have decreased, although the threat of flooding caused by ash deposition in the river continued until the very recent completion of a sediment retention dam (Perry and Lindell 1990a).

First it must be clarified that “prediction of earthquakes” means usually the estimate of time of origin, place and magnitude of an event, whereas “prediction of hazard” means assessment of the intensity of impact (ground motion) caused either by a single extreme event or by the whole family of events affecting a certain locality. The second category includes “prediction of seismicity”, i.e. the estimate of the level of future earthquake activity, e.g. in terms of the number of events of different magnitude and of the upper threshold magnitude event, during a given time interval.

The sociology of disasters has always attempted to reach a shared definition of its subject matter. The numerous critical reviews of the literature (e.g., Quarantelli and Dynes, 1977; Kreps, 1984; Drabek, 1986) identify and discuss a number of conceptualizations that are often quite diverse and sometimes incompatible. This lack of consensus shows the vitality of the discipline, which is able to keep an interest in theoretical debate, even under growing pressure to provide crude empirical data and simplified practical advice.

This paper has two aims. The first is to identify the role of communication in emergencies. The second is to stress the problems and difficulties emerging from the recent debate on these themes.

The aim of this contribution is to present some aspects of the relationship between the scientific community, the authorities, the press and the general public, regarding the prediction of possible disasters. The comments refer to the situation in Mexico and more specifically to two examples related to the seismic risk resulting from the high probability of a great earthquake in the so called “Guerrero gap”.