Earthquake Hazard and Risk

In the southern sector of the Apennines chain (Italy), destructive earthquakes with medium-high magnitude occurred within a narrow belt along the axis of the chain. Low-energy seismic swarms have frequently been observed.

The Sannio area and particularly the Benevento province, was hit by strongly destructive events of I ≥ IX MCS (1456, 1688, 1702, 1732, 1930, 1962), as well as by seismic swarms (1885, 1903, 1905).

A seismic sequence started in April 1990 near the town of Benevento. This sequence was studied in detail through the analysis of data recorded by temporary seismic networks installed in the area and operating for about two months. Over 300 microearthquakes were located in a small area of about 100 km², close to the town of Benevento. Cross sections of hypocenters indicate that the depths of the events were concentrated within the first 15 km of the crust.

The maximum seismic energy was released at the beginning of the sequence, with the highest magnitude M_L = 3.6. Most events were felt by the population, but they caused no damage.

Fault plane solutions calculated for a selected number of events show that most of the mechanisms have T axes oriented in an anti-apenninic direction (NE-SW).

In this paper a small sequence is analyzed in order to evaluate its relations with the strongest earthquakes and the tectonic lineaments associated with them, for seismic hazard assessment.

Based on the historical and instrumental data for the period 1901–1988 for earthquakes with magnitude M≥5.9 (I₀=8 MSK-64), for some main seismogenic zones and lines of Albania, in this paper we have made use of property that the time interval sequence of earthquakes follows the logarithmic normal distribution, as proved by the test of normality.

Eventually, for the different seismic regions we obtained the different time interval that the next strong earthquake might occur.

The results may help the study of seismic zoning, microzoning, seismic risk and other problems.

Statistical methods of the maximum magnitude (M_max) estimation are based usually on the assumption that the earthquake magnitudes are known exactly. Actually there are errors in earthquake magnitudes of modern and historic events. Taking into account the standard deviations of errors in earthquake magnitudes results in modified distribution of observed magnitudes. The new magnitude distribution slightly differs from the Gutenberg-Richter’s one for the large magnitudes and can explain non linear character of observed magnitude-frequency curves.

On a basis of the new distribution the formulas to obtain M_max confidence limits for large samples are derived. Numerical method for calculating exact M_max confidence limits for arbitrary sample size is also proposed.

Maximum likelihood estimates of M_max based on the new distribution are compared with the common estimates of maximum magnitude equal to the maximum of observed values. On the example of artificially generated catalogues the behaviour of the estimates for the different sample sizes and different levels of the magnitude errors is analysed. It is shown that the uncertainty in the M_max is usually much higher then the errors in initial magnitudes in catalogue.

As an example the M_max and its uncertainty was estimated for Caucasus region. We show that b value estimated with and without considering magnitude uncertainty are almost the same.

Estimates of upper bound magnitudes of seismic sources based on historical seismicity are commonly used as scaling parameters in areas that lack identified causative faults. The analysis of the earthquake catalogue of Slovenia and surrounding areas showed that for the region of Slovenia a doubly truncated exponential frequency-magnitude relationship may serve directly for this purpose, taking into account the following assumptions:

(1) The value of the upper bound magnitude, for which the corresponding least-squares estimator of the decay rate has a minimum standard error, is an optimal historical seismicity estimate of the upper bound magnitude.

(2) The difference between the estimated upper bound magnitude and the largest observed magnitude in an area with a well-defined earthquake catalogue is an appropriate increment (typically 0.1 to 0.2) for estimating upper bound magnitudes of seismic sources in the same area with the incremental technique.

There is an indication that such least-squares estimates are similar to or a bit greater than Kijko and Sellevol’s maximum likelihood estimates (1989). Both might be used with reasonable weights as alternatives for upper bound magnitudes in seismic hazard assessment in Slovenia.

To study the fundamental regularities of seismic process in the Baikal rift zone (BRZ) an attempt was made to investigate the spatial-temporal structure of seismicity of the region, using the method of fractal dimensions (Mandelbrot, 1982).

The analysis of selfsimilarity of the zone in the spatial distribution of earthquakes was made. “The completion of development” of earthquake epicentral fields of all parts of BRZ using the analysis of temporal variations of fractal dimensions in spatial distribution of events was investigated.

The resulting criteria for evaluation of “completion of development” of earthquake epicentral fields in BRZ may be used in estimations of average long-term parameters of seismic regime of the territory.

Instrumental geophysical observations in epicentral zones of strong earthquakes during aftershocks are important for a study of seismotectonic processes and a prediction of strong aftershocks. The high precision geodetic, magnetometric and gravimetric repeated measurements after a main shock have been carried out in the epicentral zones of earthquakes with M≥7.

During a gradual descent of seismic activity, the trend and oscillatoring changes of geophysical fields, exceeding background values, occur. Before strong aftershocks (K<12), near the epicenters: the rate of deformation’s change was (1–5) ppm/day, the amplitudes of local geomagnetic field variations were 7–17 nT and the gravity change was 50 mkGal.

The peculiarities of spatial-temporal change of fields reflect an unstability of deformation processes in zones of aftershocks concentration and show a character of stress redistribution at two stages of post-seismic activity: an intensive fissuring, leading to formation of blocks on a different scale (from metres to kilometres), and a consolidation of aftershock zone by smoothing of tectonic stress gradients at the block boundaries.

Recently, large catalogues of earthquake focal mechanisms have become available. The question now arises of whether the statistical methods are applicable to this type of catalogues. In this paper we discuss the problem and introduce probability density function and dispersion for focal mechanisms set, where a focal mechanism tensor is treated as a random object. The procedure of numerical simulation of random focal mechanism tensors is proposed. We also describe the process of numerical simulation of a set of random focal mechanism tensors corresponding to the given probability density function.

The GNDT seismic hazard assessment project for a future seismic zonation of the Italian territory is described and the first preliminary results are presented. They have been obtained by applying two methodologies (the Cornell approach and the mixed method), and refer to two stages of elaboration: after having summarised the products performed during a previous national project, and after the first release of the products developed ad hoc for the present GNDT project. Maps of horizontal and vertical PGA referring to 100 and 500 year return periods and maps of exceedance probability for intensity VII MCS in 50 years are shown and commented.

This contribution presents the earthquake hazard calculations that have been realized for the territory of the Czech Republic and its adjacent area. Both a review of the calculations obtained and an outlook of possible new methodological aspects in earthquake hazard calculations are summarized. The hazard calculations have been made mainly for the region of 48.5°-51.2°N, 12°-19°E, which belongs to an area of relatively low seismicity, and which is affected from time to time by moderate and strong earthquakes from surrounding seismogenic zones. These zones are located either in the Alpine-Carpathian orogenic system or on the boundaries of obviously rigid geologic blocks (e.g. the Bohemian Massif), or they are connected with neotectonic Graben systems. In order to make the earthquake hazard calculations, we delineated a large number of seismogenic zones. Influences of several factors (e.g., classification of dependent and independent events, normalization of the earthquake regime per period of observation and/or the size of the area under study, and application of different attenuation laws, effects of near-surface sedimentary deposits) upon the final hazard values have been taken into account. The earthquake hazard for the swarm pleistoseismal area of Western Bohemia / Vogtland has also been evaluated. Application of artificial intelligence techniques suggests that, in the vicinity of a resistant block of the Bohemian Massif, an earthquake-prone zone has been identified. A discussion of further steps regarding the ways in which a large amount of input data can be applied, and of an assessment of their reliability in earthquake hazard calculations is presented.

At 4 sites in Italy, a set of comparative seismic hazard estimates was performed with the aim to assess the sensitivity of each estimate to the use of different data sets. Estimates of expected average return periods for strong earthquakes have been carried out using a new approach based on probabilistic counting and logistic-type attenuation models. The different data sets used were:

the National Catalogue of Italian Earthquakes of epicentral data from 1000 until 1980 (Postpischl, 1985);

a new catalogue prepared considering all available results of recent researches on historical seismicity carried out in the framework of GNDT (National protection Against Earthquake of the National Council of Research)

data base of observed intensities compiled by the same research group.

Significant differences were found which, in particular, show the importance of observed site intensities for a reliable hazard assessment.

A recent trend in the global practice of seismic hazard assessment is the establishment of regional and global programs based on the multidisciplinary cooperation of seismologists, geologists and engineers. Here we review problems and limitations in the practice of seismic hazard assessment, the use of different probabilistic approaches in different tectonic areas and the implementation and significance of geological input in seismic hazard assessment. We also present a survey of global and regional cooperative programs in multinational and multidisciplinary seismic hazard assessment and we review the status and plan of the UN/IDNDR Global Seismic Hazard Assessment Program.

In intraplate areas of low seismicity, any possible information about the existence of a large (M_L>6.0) earthquake based on the local geology is quite valuable in estimating the true seismic hazard potential of the area of interest. Recent examples of destructive earthquakes from the seismically quiet areas such as India and Egypt, have demonstrated the often underestimated hazard potential of these areas. In conventional probabilistic hazard analyses, the seismic source zones are usually defined based on the instrumental catalogues which cover only the last 40–50 years, with a possible extension of a few hundreds of years from the historical data. The only source of information about the earlier events, is therefore dependent upon the detailed knowledge of the local geology of the area. In this study we demonstrate the importance of geological data in probabilistic seismic hazard assessments through an example from Etne in western Norway.

Detailed lineament studies conducted on satellite images, have revealed possible active fault zones in the Sunnhordland District, including the Etne Fault Zone (EFZ). The parallelism between a recently discovered NW-SE trending regional aeromagnetic lineament and the EFZ, implies that the fault zone may be related to a regional deep-seated zone of weakness which can be followed through the Precambrian basement as well as the allochthonous units. Various geological and geomorphological observations on mezoscopic and microscopic scales, as well as the positive correlation with the recent seismicity help to establish the EFZ as an active source zone. Seismic hazard map of the Sunnhordland area, where the hazard is expressed in terms of peak ground acceleration (PGA) for 10^-2, 10^-3 and 10^-4/years recurrence periods, indicate relatively high values that distinguish the seismic hazard potential of the EFZ. The contribution of EFZ to the seismic hazard estimates is clearly illustrated by the reduced PGA values obtained, when EFZ as an active seismic source zone is not included.

Earthquake Prognostics is a concept and framework for previewing future earthquakes both with repect to the natural hazard and the associated risk on man and man-made structures. It has been put into practice by implementation of artificial intelligence techniques and expert systems. In this paper, first the significance and methodology for developing for the Expert System for Earthquake Hazard Assessment abbreviated ESEHA is described; secondly, its content composition, logical structure, knowledge-base, major algorithms, case study and its results are introduced; finally some conclusion and discussion are given.

The quantitative estimation of the hazard due to future earthquakes comprises the foundation on which earthquake hazard prevetion, countermeasures and relief work are based. The loss caused by earthquakes is determined by the following two factors: a) The temporal and spatial distribution of seismic activities which is dealt with by the seismic hazard analysis and b) The degree of damage done by earthquakes on the society, economy and people is the target for the seismic vulnerability analysis. The map of the expected losses caused by earthquakes is precisely the synthetic representation of the above-mentioned two factors, and such a map has manifested the ultimate goal of seismological work to serve the economic construction and social progress.

A quantitative approach of the earthquake risk, in a long-term and large-scale sense, has been made for 23 distinct regions of Greece. The risk, Ri, in each region, i, has been determined as a convolution of earthquake hazard, buildings oldness coefficient, average population density and regional Gross Domestic Product per capita. A relative risk scale was adopted by assinging each region with R_ri = R_i/min R_i. The final product is a map of large-scale, relative earthquake risk distribution in Greece, which allows the elaboration of strategic plans for the risk mitigation.

A vast amount of information is needed in order to develop strategies for mitigating impacts of catastrophic earthquakes. This includes aspects of the natural environment, the performance of man-made structures, and the earthquake itself. Recent advances in digital data storage have made it possible to collect these data onto a single compact disk, greatly simplifying the process of data access and distribution. The authors have endeavored to collect all available sets of observational data on the Spitak earthquake of 1988. This is the first attempt to build up an information collection for a comprehensive description of a disastrous earthquake. The database has the following subdivisions: Geology, Geophysics, Main shock, Aftershocks, Impact (epicentral area, towns, buildings), Elements of Prediction, and Seismic Waveforms. Information is represented by a set of computer maps with digital data and text used as a basis for the maps. The data management system makes the data easily available to the user. This version of the database is designed for IBM PC/AT compatible computers, and includes about 180 Mb of information and data.

In July 1991, a strong motion network consisting of 8 SSA-1 accelerometric stations, was installed around the western part of gulf of Corinth, (C. Greece), in order to provide the basis for strong ground motion attenuation and site effect studies. Until now almost 550 3-component accelerograms from 130 events with epicentral distances ranging from 1 to 100 km having local Richter magnitudes ranging from 2.5 to 5.3 R, with maximum horizontal peak-to-peak acceleration values ranging from 2 to 360 milli-g, were recorded. Three of the stations were installed in the southern part of gulf of Corinth on Plio-quaternary sediments while the other five were installed on the northern part on alpine formations. The eighteen events which provided satisfactory record quality at most of the stations since 1991, have been analysed in order to define the importance of the site effects in a very geologically complex area like the gulf of Corinth.

The examination of the corrected accelerograms and corresponding response spectra at different stations shows relatively large variations in peak acceleration values as well as in frequency content at sites with different geological formations with comparable epicentral distances, underlining the importance of considering the role of site conditions in aseismic design studies.

This paper describes the results of a seismic risk study at a near-source site in Avezzano, Italy. The goal was to quantify seismic hazard through the calculation of site-dependent response spectra and ground motion time histories. The methodology used consists in combining deterministic and stochastic modeling of radiated seismic energy from an extended fault. The use of an extended source is crucial as the site is in the near-source region of the Serrone fault, which is thought to have ruptured during the 1915 Avezzano earthquake (Ms = 6.9). In a first phase of the study, we determined plausible ranges for the parameters used in characterizing possible rupture scenarios on the Serrone fault. These parameters range from the purely geometrical (i.e., fault orientation and dimensions) to kinematic (i.e., plausible hypocenters, rupture velocity, seismic moment); and were subsequently used to generate a representative family of low-frequency synthetics. The simulation of high-frequency ground motion was based on the stochastic technique of Boore (1983). Since this method implicitly assumes a point source, we developed a hybrid technique which combines Boore’s method with elements of the isochron formulation of Spudich and Frazer (1984) and Bernard and Madariaga (1984), used to generate high-frequency synthetic waveforms. The isochron formulation, as used here, provides the physical basis for the generation of realistic waveform envelopes resulting from the rupture of an extended fault. In a final step, broadband synthetics were constructed by merging deterministic low-frequency and stochastic high-frequency waveforms. Response spectra for the site of interest were then calculated, along with bounds reflecting the different rupture scenarios considered.

Taking account of the necessity for extending the frequency range to lower frequency in the seismic microzoning in an urban area, a comparative analysis of microseisms (long-period microtremors) and gravity data was carried out in the Kyoto basin, southwest Japan. The stability of spectral ratios of microseisms (soil site/rock site) was first examined through repeated simultaneous observation at soil sites and rock sites. Then spatial variation of the spectral ratios in the whole basin was mapped and compared with the characteristic feature of Bouguer gravity anomaly which was bandpass filtered using the upward-continuation technique. Main results obtained are as follows: (1) spectral features of microseisms observed both at soil sites and rock sites are remarkably affected by weather condition, but when averaged for different weather conditions, the spectral ratios (soil site/rock site) exhibit similar features, which enable us to take them as time-invariant and site-specific amplification factors; (2) the amplification factors for both horizontal and vertical components become larger stepwise from north to south in the basin, and correspondingly the frequency range having large amplification extends to lower frequency; (3) the amplification factors for horizontal components are larger than those for vertical component in the low frequency range; (4) frequency characteristics of the amplification factors for vertical component exhibit a similar shape to those for horizontal components with one-octave shift in frequency; (5) horizontal component of vibration oriented parallel to the long axis of the basin is amplified more than that oriented normal to the axis; (6) the spatial distribution of amplification factors has a close correlation with the general trend of the bandpass-filtered Bouguer gravity anomaly. These features of amplifications are interpreted from the viewpoint of Rayleigh surface waves incident to the basin with 3-D structure of bedrock, though it is suggested that some of them may be explained approximately by 1-D resonance of S and P waves in the soil sediments. It is pointed that comparative analysis of microseisms and gravity data are very useful for microzoning of an urban area with irregular configuration of bedrock, especially in the case where the information of subsurface structure is not available.

Systematic instrumental observations on the grounds with various engineering — geological properties were started in Moldova in the 1970-s. Three subsequent strong intermediate-depth Carpathian earthquakes: 4.03.1977 (M = 7.2), 30.08.1986 (M = 7.0), 30.05.1990 (M = 6.9) have also provided rich macroseismic material.

Joint analysis of instrumental and macroseismic data has revealed some peculiarities in the manifestation of different local geologico-engineering factors during the above mentioned earthquakes. Hence, it has been proposed to re-consider the existing practice of evaluation the seismic hazard in respect to this territory.