EAEE and ESC jointly organised a thematic session during XXII General Assembly of the International Union of Geodesy and Geophysics (IUGG99), within the framework of the Inter Association Symposium JSP23 on "Geophysical Hazards: Risk Assessment, Mitigation, and Warning Systems". The abstracts of all invited papers are given below.

Theme 1: THE LONG AND WINDING ROAD FROM EARTHQUAKES TO DAMAGE

D.Slejko (Osservatorio Geofisico Sperimentale, Trieste. Italy)

A.Ansal (Istanbul Technical University, Turkey)

The impact of destructive earthquakes has two faces: one in the short term and the other in the long term. The first is given by the number of victims, damaged structures and direct economic losses. The second is given by the negative influence on the social structure in the following years. Although the direct economic loses can be absorbed by the country, especially with international support, in most cases the social structure suffers permanent damages. The recent earthquakes of Northridge and Kobe have shown the long term problems caused respectively to the insurance companies and private habitants. This is one of the reasons why in seismically active countries (e. g.: Italy), a general private insurance against natural calamities is now being implemented. The knowledge on seismic risk is, therefore, fundamental for loss reduction. Seismic hazard maps at national level are the basic tool for defining the national seismic zonation which is relevant for planning adequately new settlements and constructing properly new buildings. The choice of the hazard parameter to consider for zonation is critical when the description of the whole contents of the seismic excitation is desired. However, these maps are only valid at large scales and local effects are not taken into consideration. Seismic risk maps at national scale are strategic for planning the policy for retrofitting old buildings in the presence of limited investments: their definition in a quantitative manner remains mainly a research topic for the difficulty of quantifying properly the ingredients. Risk scenarios at regional to local scale are very popular in recent years as they have the potential to limit earthquake victims and structural damage when a dangerous event is presumed to take place. In fact, based on the information and analysis concerning earthquake resistance capacity of existing buildings and structures, strengthening and retrofitting programs can be optimised. The good knowledge of site and source effects is fundamental in this case for a correct scenario definition.

Theme 2: DETERMINISTIC VS. PROBABILISTIC EARTHQUAKE HAZARDS AND RISKS

R. K. Mcguire (Risk Engineering, Inc., USA)

Deterministic vs. probabilistic approaches to assessing earthquake hazards and risks have differences, advantages, and disadvantages that often preclude the use of one over the other. Factors that influence the choice include the decision to be made (i.e. the purpose of the hazard or risk assessment), the seismic environment (whether the location is in a high, moderate, or low seismic risk region), and the scope of the assessment (a single-site risk, a multi-site risk, or risk to a region).

Decisions coming from earthquake assessments include selection of design or retrofit criteria and levels, financial planning for earthquake losses, and planning for emergency response and long-term recovery. The more quantitative the decision to be made, the more appropriate is probabilistic hazard and risk assessment.

For high seismic regions (e.g. California or Japan) where the largest earthquakes occur every 100-300 years), a deterministic scenario for the largest event will allow details to be examined such as ground motion effects caused by rupture propagation. In low seismic regions, extreme deterministic scenarios will have probabilities of occurrence that are too low to be useful for most decision purposes.

Specific site analyses generally require a probabilistic approach. Multiple-site analyses (e.g. for a portfolio of exposed or insured properties, or a lifeline) often require a probabilistic analysis because of multiple variables and complexities of the system, and a deterministic check can be misleading. Regional assessments often benefit most from deterministic models.

Theme 3: METHODOLOGICAL CONSIDERATIONS OF PROBABILISTIC SEISMIC HAZARD MAPPING

R.M.W. Musson (British Geological Survey)

The study of seismic hazard is perhaps the most practically oriented aspect of earthquake seismology. As such, it should not be treated in an idealised or academic manner, but with regard to the needs of the consumers of the final product. This has important consequences when it comes to the topic of probabilistic seismic hazard maps. Who are these for? Historically, early studies of probabilistic seismic hazard tended to be done for engineers for specific design requirements. Consequently, there has been a tendency to treat seismic hazard maps as a sort of pan-national study for engineers, who can identify the design requirements for any site by picking them from the map. A dissenting point of view argues that seismic hazard maps are by their very nature too generalised to be used in this way; that such maps provide a first indication of relative hazard and should not be a substitute for site studies. There are, therefore, a number of interesting and important methodological questions to be asked: what are the practical differences in undertaking a seismic hazard map from calculating hazard for a site? Should probabilistic seismic hazard maps have the same degree of conservatism as site studies? How can seismologists meet the needs of different audiences? An engineer may think in terms of ground acceleration, but this parameter probably means little to people in other professions who still need access to seismic hazard data, but in a form they can understand. These are questions that need to be addressed directly; one should not leave them to be answered by default.

Theme 4: EARTHQUAKE SCENARIOS FOR SWITZERLAND

Faeh, D., Bay, F., Giardini, D., Kind, F., Mayer-Rosa, D., Sellami, S. (Swiss Seismological Service, ETH Zurich); Lang, K., Bachmann, H., Wenk.,T. (Institute of Structural Engineering, ETH Zurich); Noack, T., Huggenberger, P. (Institute of Geology, University of Basel)

The goal of our project is to develop a method for the estimation of expected damage from earthquakes. The presentation is giving an overview of the state of the project, which includes, the modelling and mapping of ground motion on a regional scale for the area of Switzerland, and on a local scale for the Basel area, the classification of the vulnerability of buildings to earthquake ground motion for some target areas, and the realization of scenarios. We will present the deterministic seismic hazard in form of scenario ground motion maps. This can be done on a regional scale or on a local (microzonation) scale. On the regional scale the scenario ground motion maps should include significant earthquakes in a regional sense. Maps will then display ground motion with different probabilities of occurrence in different locations. On a local scale, ground motion scenario maps are most appropriately computed for single possible earthquakes, and these maps can be combined with a vulnerability assessment of existing structures. In this combination the scenario maps include the level and duration of shaking. Furthermore they make it possible to identify localities where ground is likely to fail through liquefaction or landslides, and they enable us to pinpoint structures that are likely to be severely damaged and to find weak links in lifeline structures. Such studies can be done with different levels of detail. The project will contribute to the necessary seismic upgrading of existing buildings, as well as to reliable earthquake resistant design of new structures, to the education of the general public, the emergency planning and it will serve as a reminder that there is a large difference between what is expected from probabilistic maps and what is a possible event.

Theme 5: REGIONAL AND LOCAL SEISMIC HAZARD ASSESSMENT

A. Marcellini, R. Daminelli, G. Franceschina & M.Pagani (Istituto di Ricerca sul Rischio Sismico)

Site effects can produce, at a local scale, abrupt changes of the ground motion as also instability phenomena such as landslides and soil liquefaction; microzonation studies are the scientific response for the evaluation and mitigation of this kind of phenomena. Seismic microzonation generally produces on a municipal scale Landuse planning criteria and defines seismic actions for engineering purposes. Seismic codes instead, are issued on the basis of Seismic Hazard studies, that is, seismic actions and seismic zonation are a direct consequence of a probabilistic seismic hazard, generally assessed considering a 474 yr R.P. In other words seismic actions assessment is biased by the inefficiency of seismic hazard evaluation procedure to account properly of parameters controlling the ground motion, mainly due to the limited number of factors used in the attenuation laws. It should be pointed out that till now this limit appears unavoidable due to the scarcity of strong motion accordingly. Microzonation investigation can overcome these limitations, but only if the reference input motion is properly assessed by adopting the same level and the same criteria of protection used to issue seismic codes at national level. The case study here presented refers to a zonation and microzonation investigation performed in the Forlì provincia (Emilia-Romagna) for the purposes of seismic prescriptions to be issued both at regional and local level and we will main focus on the importance of multiple approach for the definition of the reference input motion.

Theme 6: NEAR-FIELD GROUND MOTIONS

N. Ambraseys (Imperial College, London)

The assessment of earthquake hazard of interest to the engineer must be based on the analysis of reliable observational data than on statistics of many records and seismological parameters of questionable quality. Theoretical methods for the prediction of ground motions have become highly developed, whilst knowledge of the observational material is usually lacking. Differences between attenuation laws arise from the size and distribution of the sets of data used in their derivation and from the use of different magnitude scales, which introduce significant bias in the results. In addition, the correction of records and the modelling of attenuation laws and fitting method used to regress the data are sources of systematic errors. Taking some of these differences into consideration we find no significant variation of attenuation laws among different regions for shallow earthquakes, and a remarkable agreement between attenuation laws derived for Europe, western North America and New Zealand. They are all within the standard deviation of the determinations, which are not better than by a factor of 1.7 for accelerations. The importance of the vertical acceleration needs further investigation. However, the observation that the ratio of peak vertical zero-period or spectral value, to that of the horizontal, can be larger than 1.0 does not necessarily imply large vertical accelerations or spectral ordinates, most certainly when these maxima occur simultaneously. Also the assessment of peak and spectral ground displacements, which is of some engineering importance, needs further investigation. Near-field ground motions from medium and large magnitude earthquakes associated with surface faulting or, from sites on low-strength deposits, contain a significant component of permanent displacement, not accounted or in standard base-line correction procedures. An answer to this question is needed in order to rank these effects among other variables, and clarify the need to include or exclude them from building codes.

Theme 7: MODIFICATIONS TO SEISMIC HAZARD DUE TO SITE-CITY INTERACTION

P. BARD, P. Gueguen (Laboratoire Central des Ponts- et-Chaussies and Observatoire de Grenoble,France); J. Semblat (LCPC,France); M. Cardenas and F. Chavez- Garcia (Ciudad Universitaria, D.F)

Soil-structure interaction has long been known to significantly affect the seismic behavior of large buildings on soft soils. A few observations, and several computations as well, recently indicated the possibility for significant feed-back effects from buildings into the soil: the waves radiated back into the soil from the soil-structure interface are trapped in the surface layers when the impedance contrast at depth is large enough, so that the building is acting as a secondary source of surface waves. This phenomena has been shown to be maximum in case of coincidence between the building and ground natural frequencies: the worst case corresponds to the matching of fundamental frequencies, and induces ground motion modifications of at least 30 % in time- domain amplitudes within distances up to 5 to 10 times the building base dimensions. The question then arises of the possibility of large-scale site-city interaction due to the quasi- random superposition of such phenomena for a large number of buildings in a given city, or in a given quarter, which, in turn, raises two issues: Is the "free-field record" concept relevant in such areas? And will the construction or destruction of some buildings modify locally the hazard? The aim of the presentation is not to answer these questions, but simply to briefly review the reasons why these issues have to be addressed, and to present some preliminary, first order computations exhibiting significant effects not only for "exceptional" configurations such as Mexico City, but also for much more common situations with "ordinary" sediments and "ordinary" buildings.

Theme 8: ASSESSMENT OF STRONG EARTHQUAKE GROUND MOTIONS FOR NEAR-FAULT CONDITIONS

M. Erdik, E. Durukal (Bogazici University, Kandilli Obs. and Earthq. Res. Inst, Istanbul, Turkey)

Near-fault ground motions are strongly influenced by the earthquake faulting mechanism. Especially, the motions with periods above 1s may follow certain radiation patterns, predicted by equivalent double-couple source models, and exhibit distinct long period pulses with amplitudes depending on the orientation of the site with respect to the rupture direction. Widely use predictive earthquake engineering tools, such as empirical attenuation relationships and spectral shapes fail in the assessment of such near fault motions. Deterministic theoretical predictions of the ground motion can be achieved by convolution of the Green’s Functions and the slip function. Such deterministic predictions cannot be extended into the frequency regions above 1Hz, since, high frequency ground motions are controlled by the heterogeneities in the fault rupture, which cannot be accounted for in a deterministic manner. This requires either the use of stochastic source models or the stochastic treatment of the high frequency components in the ground motions.

Based on these developments a state-of-the art hybrid procedure is developed for the assessment of the time history of the DBE (or SEE) ground motion for important engineering structures near major faults. The essential elements of the procedure can be listed as follows: (1) Assessment of the source parameters of the DBE motion associated with the corresponding return period for specific conditions of site and seismicity; (2) Deterministic assessment of the low frequency (DC-1Hz) ground motion, at the outcrop of a reference soil layer, due to rupture of seismic faults using numerical simulation; (3) Use of a Boore (1983) type stochastic simulation method to complement the deterministic low frequency ground motion with high frequency (1Hz- 50Hz) components; (4) Combination of the two parts of ground motion to yield a site-specific simulation for a frequency range of DC-50 Hz.; and (5) Site response analysis, if required, to include the local wave propagation effects in the soil media above the reference soil layer. An example that illustrates the application of this procedure is provided.

Theme 9: LOCAL SITE EFFECTS AND DYNAMIC SOIL BEHAVIOUR

E. SAFAK (U.S. Geological Survey)

Amplitudes of seismic waves increase significantly as they pass through soft soil layers near the earth's surface. This phenomenon, commonly known as site amplification, is a major factor influencing the amount of damage on structures. It is crucial that site amplification is accounted for when designing structures on soft soil sites. The characteristics of site amplification, at a given site, can be estimated by analytical models, as well as field tests. Analytical models require as inputs the geometry of all soil layers from surface to bedrock, their dynamic properties (e.g., density, wave velocity, damping), and the incident bedrock motions. Field tests involve recording and analyzing the dynamic response of sites to artificial excitations, ambient forces, and actual earthquakes. The most reliable estimates of site amplification are obtained by analyzing the recorded motions of the site during earthquakes.

This paper presents a review of the models and methods that are used to characterise site amplification, and introduces some new ones with better theoretical foundations. The models and methods discussed include spectral and cross-spectral ratios, response spectral ratios, constant amplification factors, parametric models, physical models, time-varying filters, methods for downhole records, single-station methods, and generalised inversion techniques. The paper also examines the validity of one-dimensional models, in comparison to 2D and 3D models, and shows the effects of surface waves and surface-to-bedrock topography on site amplification estimates. The paper concludes that probabilistically cross-spectral ratios give more reliable estimates of site amplification than spectral ratios. Spectral ratios should not be used to determine site amplification from downhole-surface recording pairs because of the feedback in the downhole sensor. Response spectral ratios are appropriate for low frequencies, but overestimate the amplification at high frequencies. One-dimensional site amplification models are not appropriate for sites in deep sedimentary basins that are susceptible to generating surface waves. The surface topography, as well as the topography of below-surface layers, cause focusing and defocusing of seismic waves, which can create drastic changes in site amplification in very short distances.

THE CYCLIC BEHAVIOUR OF SOILS AND EFFECTS OF GEOTECHNICAL FACTORS IN MICROZONATION

A. Ansal (Istanbul Technical University, Turkey)

The behaviour of soils subjected to cyclic loading is briefly reviewed. The results obtained from cyclic triaxial, simple shear and torsional hollow cylinder tests conducted on undisturbed, normally consolidated clay samples subjected to different shear stress amplitudes and different loading patterns are summarised. Results of the tests conducted under uniform cyclic stresses to evaluate "cyclic yield stress" and "threshold cyclic shear stress" are presented. The effects of cyclic loading on static shear strength and induced settlements due to pore water pressure dissipation are discussed. Cyclic tests performed to evaluate the liquefaction susceptibility of laboratory prepared and undisturbed sands and silty clayey sands are briefly reviewed in the light of the findings reported in the literature.

Geotechnical site conditions is one of the main factors controlling earthquake forces affecting structures. Therefore, in analysing the observed damage in previous earthquakes and for microzonation studies, this factor and its coupled effect with earthquake source characteristics need to be evaluated. The earthquake source characteristics induced by a tectonic source mechanism are on macro level and are not sufficient to explain the variations in structural damage observed within short distances. On the contrary, the geotechnical site conditions that can be very different due to changes in the thickness and properties of soil layers, depth of bedrock and water table, can have more dominant influence on damage distribution. In addition the effect of coupling between site and source characteristics may modify earthquake ground motion characteristics significantly. There are large numbers of instrumental field observations obtained during recent earthquakes reflecting the coupled effects of geotechnical site and earthquake source characteristics. During earthquakes soil layers are subjected to multi-directional cyclic stresses with different amplitudes and frequencies that lead to cyclic deformations and to changes in stress-strain and strength properties of soil layers. Extensive laboratory, model and field studies were conducted concerning response of soils subjected to cyclic stresses. Significant improvements were achieved in the field of insitu tests to obtain more reliable soil properties. Numerous analytical and empirical relationships were developed to model the behaviour of soil deposits subjected to earthquake excitations. From an engineering perspective, it is possible to investigate the properties of geo-technical site conditions in detail and analyse the response of soil layers with sufficient accuracy.

Theme 10: EVALUATION OF VULNERABILITY OF CIVIL ENGINEERING STRUCTURES

M. DOLCE (University of Basilicata, Italy)

The impact of earthquakes on man-built systems (buildings, groups of buildings, lifelines, cities, etc.) is usually expressed in terms of losses. Loss assessment is a highly complicated task, involving, besides structural engineering and seismology, many disciplines such as geotechnical, transport, hydraulic and electrical engineering, geology, urban planning, social and economic sciences, etc. However most of the losses results from the consequences of direct damage to civil engineering constructions, particularly buildings. The evaluation of their seismic vulnerability is then a fundamental step in the process of determining the impact of earthquakes on man-built systems. This is usually assessed in terms of seismic risk in a period of time (e.g. one year, hundred years), i.e. in terms of probability or of expectation of losses during that period due to all the possible arriving earthquakes. This representation is very general but presents many drawbacks such as the difficulty of interpreting the results in practical terms and the difficulty of expressing and quantifying losses for territorial systems. In fact they are highly dynamical systems, whose future developments are difficult to forecast, particularly after the occurrence of an earthquake. A long term loss prediction is therefore not much significant, if it is referred to the current situation or even to the current trend.

Recently the interest of researchers and operators has been focused on risk scenarios, where the impact of a given earthquake is investigated and quantified. This approach permit to better understand the behavior of the built environment under study and to take the countermeasures aimed at reducing its impact. Scenarios can be prepared considering different aspects of the earthquake consequences. Different levels of accuracy can be assumed in the preparation of a scenario, starting from the assessment of shaking intensity and characteristics, through damage prediction up to loss assessment. Obviously the accuracy levels in the different steps must be consistent.

The assessment of damage to constructions deserves a special attention. In this respect, two ingredients are fundamental to prepare well-grounded damage scenarios: the vulnerability evaluation and the inventory of the man-built system components. There must be a strong relationship between the accuracy of the vulnerability evaluation and the details of the inventory of constructions whose vulnerability must be assessed.

Theme 11: FROM VULNERABILITY OF OBJECTS TO VULNERABILITY OF SYSTEMS

C. GAVARINI (Universita' La Sapienza, Roma, Italy)

First of all the paper outlines the current conceptions relevant to vulnerability of constructions, describing the various approaches and the different research levels, then, in the second part, a more global vision is developed in which the constructions belonging to an aggregate, or a centre, or a city, or a territorial area, are considered as a part of a vulnerable system, with all the complexity that it brings about, in terms of different variables that must be taken in consideration, properties and values that are in danger, disciplinary, cultural and historical aspects that must not be ignored. Assessment of vulnerability is strictly connected with another basic problem, risk mitigation, that will be briefly considered in the third part, analysing the interrelations between vulnerability and environment aggressions. Also here we must point out how today it is a common exigency, also as regards technical codes, to abandon the sectional vision of objects considered separately and to promote instead the above mentioned ‘picture in picture’ vision applied to systems, better said to complex systems, specially when the areas in question are rich with particular ‘objects’ with a ‘cultural value’, such as historical buildings and monuments. Lastly, in the fourth part, we return to deal with single objects, giving a short account of problems specific to monuments and in particular to churches subjected to seismic actions; referring to studies regarding the Cathedral of Noto, partially collapsed in 1996 and now in course of reconstruction. The need for experiments, either real or virtual, is recognised as a key for understanding and classifying the…

Theme 12: LOSS ESTIMATION: A POWERFUL TOOL FOR RISK ASSESSMENT AND MANAGEMENT

F. Bendimerad (Risk Management Solutions, Inc, USA)

Earlier loss estimation studies were limited to investigating particular scenarios and were carried out by highly specialised experts. Today, loss estimation techniques are translated into efficient software applications that are accessible by a large constituency of end-users. These techniques offer a high level of analysis sophistication and enable users to perform various "if-then" scenarios to study the sensitivity of the results, to develop a better understanding of the outcomes and to gain insight on the consequences of the findings and decisions.

Loss estimation techniques have benefited from the advances in information technology. Modularity, encapsulation and a new generation of computer codes such as C++, provide a logical and flexible structure for organising the analytical procedures involved in loss estimation. These techniques organise the multitude of analytical calculations into modules that are logically inter-related by hierarchical rules. This flexible architecture permits ease in development, testing, validation, maintenance and upgrade. The study region is divided into geographical units (Geo-Units) such as postal codes or census tracks. The data, calculations and results are then associated with the centroid of the Geo-Unit. The aggregated results from the Geo-Units yield the results for the study region. Geographical Information Systems (GIS) and Relational Database Management Systems (RDMS) are used to organise data in data warehouses, to manipulate data during analysis, and to associate results to geographical regions from which they can be queried, aggregated and/or mapped. GIS and RDMS also allow for easy display of input and output (in standard reports and maps) providing a critical functionality for communication of outcomes to end-users. Loss estimation has become a critical tool to the insurance industry and is quickly being adopted by a wide range of users including emergency managers and planners. The key to this expansion is the integration of new information technologies that gave these techniques greater analytical capabilities, flexible architecture, and user-friendliness.

GLOBAL SEISMIC HAZARD AND RISK ASSESSMENT

Y. Chen, Q. Chen, L. Chen, J. Li and J. Liu (Seismological Bureau, China)

A global seismic hazard assessment was conducted using the probabilistic approach in conjunction with a modified means of evaluating the seismicity parameters. This method is applicable to both oceanic and continental regions, and for any specific duration of time. It can be used for those regions without detailed geological information or where the relation between existing faults and earthquake occurrence is not clear.

Most seismic risk studies use a probabilistic approach in which predicted damages in various categories of structure and facilities in the region in concern are estimated and added together to obtain a total loss for particular intensity ranges. We have used an alternative means of estimating earthquake losses based on several macroeconomic indices such as the gross domestic product (GDP) and population. A global seismic loss map is then compiled.

The expected losses (in USD) of selected countries and regions for the next 50 years are listed bellow:

Region Loss (USD) Region Loss(USD)

World 949 Asia 563

N.America 15 S.America 60

Japan 390 USA 66

China 17 Europe 184

Oceania 12

Combination of the various ingredients by GIS techniques, quality definition of the ingredients used, etc.