Natural Hazards 14: 85-88, 1996. (~) 1996 Kluwer Academic Publishers. Printed in the Netherlands.

85

Book Review Alberto Carrara and Fausto Guzzetti (eds.): Geographical Information Systems in Assessing Natural Hazards, Selected Proceedings of the Intemational Workshop on Geographical Information Systems in Assessing Natural Hazards, Perugia, 20--22 September 1993, Advances in Natural and Technological Hazards Research, Vol. 5, Kluwer Acad. Publ., Dordrecht, 1995. Kluwer has achieved another first. This time we are presented with the first comprehensive book applications of geographical information systems (GIS) to the assessment and prediction of the effects of natural hazards. This is a selection of 16 papers representing slightly over one-third of the contributions given at the 1993 Perugia Workshop. The editors warn us that GIS has become a mixed bag of techniques used by a wide range of disciplines for "displaying data in a fashionable way". On the other hand, no consistent set of analytical procedures has emerged among geographers for predicting the effects of natural hazards in space and time. There are several reasons for this, not the least of which is the incipient state of fundamental research on the social effects of natural disasters. Is this situation 'unhealthy'? In other words, is our baby sick (as the editors believe), or is it merely small?

1. Some Applications The question is complex. It is addressed from different perspectives by two geographers, D. E. Alexander and J. Terry Coppock. But before going into their analysis, let us consider some of the applications of GIS which make up the substance of the book. These case studies are very instructive. They deal with GIS as applied to floods, landslides, earthquakes, volcanic eruptions and groundwater pollution. Several authors compare GIS to a 'toolbox', a recurrent image which stresses the emphasis on technology as applied to natural hazards. I was much impressed by Detti and Pasqui's paper describing a computer program which generates drainage patterns from digital elevation data. The proposed technique surprised me. It seemed very complicated. Wouldn't it be easier to solve the problem by simulating rainfall and runoff with some cellular automata approach, and allowing the given topography to find its own watersheds? One wonders whether this has been tried.

86

BOOKREWEW

The five papers on slope instabilities are most instructive. A critical review paper by Carrara, Cardinali, Guzzetti and Reichenbach discusses 'errors' (i.e., lack of consistency) between different predictions using the same method. Experiments conducted in several different areas suggested that these 'errors' were extremely high. For example, in the La Honda, California drainage basin, the landslide predictions by two equally experienced surveyors overlapped by less than 10%. More than 20% of the total area had been classified as unstable by one or the other surveyor. If enough experts are used, it seems likely that the entire area will eventually be classified as hazardous! Some of the reasons for this inconsistency become clear by reading the other case histories. The assumptions and methods vary considerably. It would seem that one way to constrain the assumptions and improve the evaluation of errors might be to compare the predictions directly with the actual occurrence of events. But this approach does not seem to be very popular. There is an inherent jinx in technology which causes it to expand in proportion to its ineffectiveness. In the case of GIS, the technology may tend to become more important than its relevance to the actual events to be predicted. Most authors in this volume seem to be aware of this limitation. Thus, Lanza and Siccardi discuss the use of GIS-type systems for flood forecasting in terms of actual flood monitoring. In other words, this application is promising because it provides a link between the prediction and the phenomenon itself. Here is a lesson worth retaining: any technology, in order to be capable of forecasting a geophysical event, should at least be able to monitor the phenomenon to be forecast. At least one very elaborate GIS project seemed to this reader to lack a component of verification. The prediction of geological hazards in San Mateo County, California was begun after 1968 (though actual GIS techniques are being applied since 1986), by an interdisciplinary group of USGS scientists. A large number of hazard maps was produced over the years, using a variety of different approaches. Yet the extensive article on the San Mateo project fails to mention the 1989 Loma Prieta earthquake which evidently put these predictions to the test. Most of the maps included in the article seem to support the predictions; but a detailed confrontation with the 1989 earthquake was apparently not perceived as particularly relevant. Could it be that the predictors were so absorbed by their task that they lost sight of the real phenomenon to be predicted? A comment by the author on this point would have been helpful.

2. The Vagaries of Theory Let us now consider the two review papers which introduce the volume. The survey by David E. Alexander reflects both the best and the worst in the field of disaster studies. The author highlights his own contributions which are very illuminating, though not always in the way they were intended. There is the perverse insistence of

BOOKREVmW

87

linearizing what is essentially a nonlinear problem. Thus, the equation which purports to give the 'human impact' of a geophysical event cannot possibly describe a disaster (which is a discontinuity), because it is linear. The work of Gilbert White and the Chicago school of geographers is quite properly given first billing, though regretfully its connection with the work of Pitirim Sorokin is not acknowledged. This spells trouble, for the overall account of social disaster research provided in this paper tends to be unsatisfactory and shallow. For example, Alexander castigates the "practitioner's reluctance to undertake interdisciplinary work" and feels strongly about the "siege mentality in the social sciences", which he attributes to "the dismissive or uncomprehending attitudes that many physical scientists have toward social studies". As a physical scientist, I would reply that it won't do to blame our know-nothing attitude (which, I am afraid, is bad enough as it is), when the scholarship of the social scientists is made to look puny and unappetizing. I think the review is unfair to the sociologists. Honest critics like Hewitt are only mentioned in passing and European scholars such as Wolf Dombrowsky not at all. Sociology, it should be recalled, was bom as a critical analysis of modem society and cannot be reduced to a collection of recipes for evaluating the 'perception' of reality by human actors. Predictably, Alexander quotes approvingly from Quarantelli that "social problems can only be dealt with socially; technological improvements can only address technological problems". Presumably, technology originated from outer space, not from within society. Coppock's survey paper is limited to 'a GIS perspective', which may be just as well. His introduction refreshingly recognizes that the IDNDR emphasizes technology because of the dominant role played by developed nations (who wish to sell technology such as GIS) in proposing and promoting the Decade. The trouble seems to be that there are various computer packages for GIS applications but no single connecting technique with hazard studies. This has not yet been developed. The lack of a solid basis in theory is particularly difficult to overcome. The paradigm of impact-vulnerability originated by White and his followers is hard to translate into spatial patterns. For what is vulnerability? Isn't it just another word for 'hazard', which is precisely the information we are after? The conceptual difficulties loom so large that Coppock tends to give up and confuse population density with vulnerability. He claims that control of natural hazards "is best done by measures to reduce the population at risk". At worst this is neoMalthusianism and at best, a recipe for inaction. After all, the hazards, if left alone, might take care of reducing the population without further assistance by experts! Loose ends in theory are all too evident. A natural disaster is defined as 'the occurrence of a natural hazard on a large scale'. More commonly, however, hazards are not considered events but probabilin'es of occurrence of potentially damaging events. Fuzziness of theoretical concepts may be damaging to progress in technological applications such as GIS. In the end, the author's attempted distinction between disaster prevention in developed and developing countries boils down to

88

BOOKREVIEW

a recommendation for developing countries to 'keep it simple', and to strive for increased centralization. In my experience, the opposite is true. The less 'developed' (i.e., industrialized) a society, the more sophisticated its technology needs to be in order to cope with disaster management. This makes sense, because the society has not yet become homogenized through industrialization. As for centralization, if it works at all it would be in an industrialized nation like the U.S. or Japan, where the fragmentation of emergency services has already gone too far. On the whole, it may surprise the reader that elementary 'lessons' of past disasters have not been more widely assimilated. This is a most interesting and stimulating volume. It shows what happens when a potentially useful and enlightening technique is brought to bear on a vital but exceedingly difficult problem. The book should be of interest to all specialists in natural hazards.

National University of Mexico

CINNA LOMNITZ