1. Introduction

Strong-motion instrumentation and recording in Europe and in the Middle East started much later than in United States and Japan, and developed slowly, chiefly with analogue instruments. Table 1 lists the year of the first recording made in various countries in the region since 1967. With the advent of digital recorders in recent years, this development increased rapidly, particularly because of the need to instrument major engineering works and public buildings and to comply with the requirements of hazard assessment and earthquake resistant design stipulated in EUROCODE-8.

2. Strong-motion networks

A recent survey made by the Working Group on Strong-motion Studies (TG2) of the European Association for Earthquake Engineering (EAEE) to assess the present state of strong-motion recording capabilities in the region, shows that although the total number of all the stations is difficult to estimate, the number of instruments operating in the free-field is close to 1,500 (Table 1). That we are aware of, the number of individual triaxial recordings made by earthquakes of all magnitudes during the last 28 years exceeds a conservative estimate of 2,500. Appendix 1 lists the agencies responsible for the operation of strong-motion networks and/or of data banks in Europe. This list does not include statistics from the former USSR and from a few other European countries, or from the European nuclear and oil industries.

The survey also shows that European strong-motion networks and individual stations have been established and are maintained with recurrent government subsidies or short term grants. They operate as independent state, industrial or university units with little or no coordination between them. Some of these networks are very well run but because of the closed system within which they operate, even within the same country, few cooperative research programmes have developed between them and only a fraction of their output reaches end-users, engineers and earth scientists alike.

3. Research programmes

Of the research programmes, the following are in progress: Arrays and dense networks to investigate path and soil amplification effects at:

• Volvi (Greece) biaxial network installed in 1994, (VOLV).
• Valle del Sacco (Italy) biaxial array installed in May 1984 (ENEA).
• Umbria (Italy) biaxial array in operation since 1992 (ENEA).
• Aterno Valley (Italy) triaxial array put in operation in early 1996 (SSN).
• Friuli (Italy) biaxial array (UT).
• Ohrid Lake, (Macedonia) triaxial array in operation since 1989 (OH).
• Granada Basin (Spain) array installed in 1996 (GB)

Table 1. List of strong motion instruments

#: First recording by UNESCO missions

Note: To the best of our knowledge the number of instruments shown are in free-field conditions unless marked with an asterisk.

4. Strong-motion data centres

Data from other strong-motion centres still are not publicly available. However a number of data centre provides information on the Internet about the strong motion records archived in their databank and/or data of their accelerograph networks:

• seismological algorithms.
• earthquake slide sets.
• the worldwide earthquake database (more than 1,75 million earthquakes).
• the significant earthquake database (more than 5,000 earthquakes).
• the earthquake intensity database.
• seismograph station bulletins database.
• world-wide strong motion data archive (more than 15,000 digitised and processed records from 1933-1994)

This INTERNET sites provides newly processed strong motion data from very recent Californian earthquakes,

Browses, finds, views and plots accelerograms (time series and spectra) from the CALTECH and other databases. Also computes and plots inter-story demand spectrum,

Access to acceleration records from the Guerrero accelerograph network (more than 30 digital strong motion accelerographs in Guerrero and neighbouring states in Mexico),

5. TG2 work programme

This programme was initiated by the European Association for Earthquake Engineering and aims at three work areas:

• the establishment of a freely available data base and bank of free-field strong-motion data (records and parameters).
• the study of source, path and local effects.
• the study of structural response and establishment of an available data base and data bank of structural strong-motion data.

6. Attenuation laws

A number of peak ground acceleration attenuation laws have been published for Europe by Chiaruttini and Siro (1981), Sabetta and Pugliese (1987), Petrovski and Marcellini (1988), Ambraseys and Bommer (1991), Theodulidis and Papazachos (1992), Tento, Franceschina and Marcellini (1992), and Ambraseys (1995). Spectral ordinate relations have been published by Petrovski and Marcellini (1988), Pugliese and Sabetta (1989), Mohammadioun (1991), Caillot and Bard (1993), Lee (1995), Ambraseys, Simpson and Bommer (1996), and Ambraseys and Simpson (1996). A discussion of some of these laws is given by Ambraseys and Bommer (1995).

7. Discussion

During the past decade Imperial College (IC) / ENEL (Rome) / ENEA (Rome) / IPSN (Paris), the Tripartite Project Group (TPG), attempted to retrieve, process and analyse strong-motion records, chiefly analogue, from the European and Middle Eastern regions and to identify future needs for strong-motion information. In what follows we describe briefly the method we followed to retrieve and process these data and the lessons learnt from this exercise.

For the engineer, analysis of existing strong-motion recordings is the most common method used to estimate future ground shaking in a seismic hazard analysis. This method must rely on good quality databases, uniformly processed records supported by reliable seismological and soil mechanics information and reliable associated data banks.

Most of the TPG records come from events before 1995, and are in the period already analysed by the International Seismological Centre (ISC) and Harvard. Many of the earthquakes are of moderate magnitude and are reported from a relatively large enough number of stations to ensure reasonable azimuthal coverage. The locations found by ISC are therefore not likely to be in serious formal error and can be used as initial values for refinement. The main uncertainty is in depth of focus, and adjusting this could introduce compensating changes in position, particularly if the reporting stations are concentrated in a limited range of azimuth. For well-recorded events this effect is likely to be small, unless there are strong lateral velocity inhomogeneities, again unlikely in this area. The problem of depth may not be all that important in California but for certain parts of Europe and for other areas that are of interest to us, such as Central America for instance, focal depth is an important consideration.

Teleseismic locations are known to have larger uncertainties compared with those from local networks and the latter, when derived from special studies have been adopted in our analysis. Like epicentres, focal depths are also based on teleseismic arrival times alone and lack precision. Here again, results from special studies have been incorporated into our analysis to improve the quality of the data. Some of our records were generated by lower crust or subcrustal earthquakes and these have been treated separately.

The use of a unified magnitude scale in attenuation studies is an important consideration. Our adoption of Ms rather than ML stems from the fact that the former is not only the best estimator of the size of a crustal earthquake, but also because seismicity in Europe is generally evaluated in terms of Ms, it is necessary to use the same magnitude scale in attenuation relations. Moreover, we have chosen Ms because in some parts of the study area there are no ML determinations. Also, importantly, because we have access to a large number of station bulletins we can calculate Ms uniformly for almost all events that have generated strong-motion records from a sufficiently large number of stations using the Prague formula, which was checked for Europe (Ambraseys and Free, 1997). Equally important in the assessment of reliable source distance, particularly in the near field, is the exact location of the recording sites. This requirement is obvious and it may sound superfluous to check the station coordinates supplied by local networks. However, the re-examination of a number of files showed location errors by certain agencies, due to systematic reading errors or misprints up to 12 km. The solution to this problem has been the relocation of stations using hand-held GPS instruments.

The distance or source-path one assigns to a strong-motion record has a significant influence on the close-in behaviour of attenuation curves, particularly for small events for which location errors can be many times the source dimension. These errors accrue owing to errors in source and station locations. For most of the larger earthquakes we adopted the closest distance to the projection of the fault rupture. For small magnitude crustal events the source distance are close to the epicentral distance. However, the locations of some of the smaller events are poorly known and for this reason their location were re-evaluated or distances based on S-start times (first S-wave arrival-trigger time) were adopted.

Local site conditions (soil, topography, instrument location, housing and characteristics) at many European strong-motion stations are poorly known, particularly for the case of old sites which have been moved or abandoned, or for temporary stations. In terms of the soil conditions, the majority of sites can only be described at best in very general terms such as “soil” or “rock”. There are however, some stations for which there is no knowledge of the soil conditions. For a small percentage of European stations, relatively detailed descriptions of the local soil conditions do exist. These typically take the form of shear- and compressional-wave velocity profiles, although some are described in terms of Standard Penetration Test (SPT), void ratios, density, moisture content and other material property data. For these sites, velocity values can be estimated roughly via empirical relations.

The topographical details at most stations are even less well described. Where they do exist they may be given only in terms of very broad descriptions such as “at the top of a hill” without any reference to the hill dimensions or the surrounding geomorphology.

Instrument data is usually more readily available, at least in general terms of the instrument type and the structure in which it is housed. However, it is not uncommon to have no knowledge of the specific characteristics of the instrument (sensitivity and damping). Furthermore it is even less common to have detailed information regarding the structure in which the instrument is housed, for example building plans, and the specific location of the instrument within the structure beyond, for example, “the basement of a 3-storey structure”.

The TPG data bank and database contains information for Europe and worldwide regions. Specifically for Europe, the database contains information for 869 separate events, which are believed to be associated with strong-motion recordings. The completeness of data for these 869 events is variable ranging from only the basic event time to a complete set of magnitude values, moment, stress drop, fault type etc. The chronological distribution of these 869 events is shown in Table 2.

Table 2: Distribution by country and year of earthquakes, which are believed to be associated with strong-motion recordings.

Afg: Afghanistan, Alb: Albania, Alg: Algeria, Bul: Bulgaria, Ger: Germany, Gre: Greece, Ice: Iceland, Isr: Israel, Nor: Norway, Pak: Pakistan, Por: Portugal, Rom: Romania, Tur: Turkey, UK: United Kingdom, USSR: Former Soviet Republic, Yug: Former Yugoslavia

Of the 869 events, 617 are known to have triggered strong-motion instruments. Information for 1249 triggered record sets from these 617 earthquakes is contained in the database. As previously, the completeness of information is variable ranging from basic identification of the triggered station through to peak acceleration values, intensity levels, source-site distances, instrument type, basic soil condition, ground or structural record etc. The distribution of these events and triggered stations is given in Table 3. From the 1249 known triggered record sets, 711 record sets from 331 separate earthquakes, giving a total of 2293 component records are held on the Imperial College data bank. The 711 triggered sets comprise 682 three component recordings, 20 two component recordings, and 9 single component records. The 2293 records include 195 duplicates resulting from multiple digitisation of records by the same or different agencies. The distribution of the 331 events and 711 record sets are shown in Table 4. Uniform surface-wave magnitude determination and distance calculations have been done where possible for these 331 events and 711 record sets.

The 2293 records in the data bank come from 261 individual strong-motion stations. Of these 261 sites, there are 19 for which there is no knowledge of the local soil conditions. For the other 242 stations, 182 can only be described in general terms following the classification scheme of Boore et al. (1993) which considers sites in one of four categories based on an average travel-time velocity to 30m depth, V_t30. For the remaining 60 stations, shear-wave velocity profiles are available, or can be estimated from SPT N-count and void ratio data. Of the 60 sites, local soil data is available at or very close to the instrument for 47 stations. For the remaining 13 sites, local soil data has been extrapolated from nearby stations or from general profiles which are considered representative for the area of concern.

Of the 261 stations associated with strong-motion records, 11 are considered to be structural. These 11 stations recorded 50 separate record sets giving 148 individual components. Most interest though at IC and from the TPG has been only with ground level records. Structural records have therefore been excluded from analyses. The resulting available ground-level data set is therefore comprised of 2145 component records (661 record sets) generated by 321 separate earthquakes and recorded by 250 different ground-level stations.

Table 3: Distribution by country and year of earthquakes, which are known to have triggered strong-motion instruments (Ev) and the associated number of triggered stations (St).