Observatories and Research Facilities for European Seismology

Broadband seismographs may be classified according to the range of seismic signals they can record in a single channel. We propose as a minimum requirement for a broadband seismograph that it must record all seismic signaks that can be recovered from standard (WWSSN) short- and long period seismograms. As a development goal we propose the ability to record all teleseismic signals from the level of ground noise to the amplitudes encountered after a magnitude 9 earthquake at 20 deg. distance. This type of instrument is specified in the design goals of the IRIS global network and will be called a very broadband, or VBB, seismograph.

1.1 The current plans for a Federation of digital broadband seismic networks bring up the problem of standardization. While a standardization of digital data formats is clearly desirable, it is not so clear what consequences an attempt to specify the performance of the hardware would have. Standardization in this field may have undesired commercial implications. Also, the geographical situation of a station is in most cases more important than its technical perfection, and we certainly don't want to exclude any useful data for formal reasons. We must, however, define what we mean with broadband data, and also try to find a common goal for future developments in order to maintain compatibility in a time of rapid technical progress.

1.2 The historical subdivision of the seismic spectrum into a short-period and a long-period band is due to the presence of marine microseisms at intermediate periods (2 to 8 sec) which obscure visibly recorded seismograms unless they are substantially attenuated in the seismograph. Although it can be shown that no useful information in the intermediate band is suppressed by the response of a conventional SP/LP seismograph combination, the limited resolution of photographic records makes it difficult or impossible to retrieve this information.

Digital recording has eliminated this problem and has permitted to record short-period and long-period signals on the same medium. Signal components in the intermediate band can now be evaluated as conveniently as those at other periods. The broadband technique gives access to seismic waves whose length is in order of 10 to 50 km, a range that is of highest interest both for the study of earthquake foci and for structural investigations.

Although broadband seismic systems can be realized with analog magnetic recording, that techique is now obsolete and we will henceforth assume that the broadband data are recorded in dogital form.

2.1 As a minimum requirement we propose that a broadband seismograph must resolve all seismic signals than can be recorded from standard SP and LP seismograms. Broadband data must permit simulation of all conventional SP and LP responses, as shown for the GRF array by D.Seidl in J.Geophys. 48, 84-93, 1980.

2.2 Broadband recording in this sense implies a passband at least from 0.5 to 100 sec, and a response in that band that is sufficiently simple, stable and well known to be removed from the data. Different technical realizations of the concept are possible; the German GRF array and the Dutch NARS array may serve as examples for observatory-type and mobile broadband installations. For compatibility with the GRF data, it is desirable to use a rate of 20 Hz for data exchange, although the original sampling rate might be higher. In fact, it is an advantage to sample originally at a higher rate and decimate the time series after digital low-pass filtration because in this case almost the full Nyquist bandwidth is available for the data.

Continuous recording might be too costly or impractical for many potential stations. The proposed minimum requirements can also be translated into trigger criteria for event recording; this task is however beyond the scope of this draft.

3.1 The design goals for a new Global Seismographic Network (IRIS 1985) define a broadband system that can record seismic signals in a frequency band from the shortest teleseismic periods ( ~ 0.1 sec) to the free oscillations of the Earth ( ~ 1 hour), and over a dynamic range from ground noise to signals from a magnitude 9+ earthquake at 20 deg. distance. Station sof this type will be called very-broadband or VBB in distinction to the minimum broadband or BB type. Since a number of experimental VBB stations with slightly relaxed specifications are already in operation and have proved their performance, the disign goals are certainly realistic. Very-broadband data can replace all other teleseismic data; VBB recording therefore seems to be the most economic method of digital seismic recording.

Since the essence of the VBB concept is simply to record all teleseismic signals, VBB recording appears as a natural goal for all networks devoted to regional and global seismology.

A seismometer that resolves ground noise over the entire teleseismic band and has a dynamic range of 140 dB or more at long periods
A digitizer with an adequate dynamic range and a good lianarity
A station precessor that can be reprogrammed in a higher programming language and can be interfaced to future recording devices and telecommumication equipment
Digital recorders of any kind kind with a capacity in the order of 100 MByte or more
Auxilary components like a buffered power supply, a precise system clock, a time-signal receiver, a status monitor etc. which are however part on any self-contained seismic system and will not be discussed here

We will not attempt to give technical specifications for all components; an outline of the main requirements and of the state of the art may however be helpful.

3.2.1 Seismometer The resolution and linearity required for VBB seismic recording can at present only be achieved with small, hermethically sealed, and well- isolated (or borehole mounted) instruments with negative force feedback. TYhe sensor response must be carefully chosen in order to minimize the required performance of the digitizer. A response that is flat to velocity over a large part of the VBB band can be combined with a digitizer of conventional design; a response that is flat to accelaration is easier to real;ize but requires a digitizer of extremely high performance.

For surface installations, a suitable seismometer is now available (the Streceisen STS-1 VBB feedback seismometer). It has a response flat to ground velocity from 0.1 to 360 sec and resolves ground noise to periods as long as one hour. Teledyne-Geotech is developing a VBB borehole seismometer , the TG-540000, on the basis of the KS-36000 system used in the SRO network. Specifications for this instrument are available; a prototype is being tested at the Albuquerque Seismological Laboratory of the USGS. The Guralp CMG-3 seismometer which is offered in surface, posthole and downhole packages appears to be in the BB rather than in the VBB category.

3.2.2 Digitizer The digitizer is a critical part in a VBB seismograph. It must however cover a dynamic range of at least 140 dB, and it must have an excellent linearity and resolution because signals of greatly differnt amplitude may simultaneaously be present in different parts of the VBB seismic band. Gain-ranging digitizers can easily provide the required dynamic range but are notorious for nonlinear distortions. if such a digitizer is used, care must be taken that the fundamental range is large enough so that microseisms and local noise do not force gainranging. Even than, nonlinear distortion of larger earthquake signals may occasionally be noticable.

A perfect solution to the problem of digitizing VBB seismoc signals has recently emerged in the form of fixed-range, 24-bit digitizers. Such a digitizer is available from Gould; an alternative disign has been developed by J.M.Steim of Harvard University and others. Although the price of the Gould digitizer is still high due to the high costs of development, it is remarkable that the basic design of the new digitizers is quite simple and no expensive components need to be used. Within a few years, 24-bit digitizers should become less expensive than present gainranging systems, and still offer a linearity and resolution of magnitude better than these.

3.2.3 Station processor A station processor is not strictly an essential part of a digital seismograph, yet seismic data acquisition systems are normally designed around a microcomputer for greater flexibility. The rapid developments in the field of starage media and telecommunications would otherwise make the system obsolete in a few years. Also, the amount of data generated by a VBB seismic system is considerable; a station processor incorporating data compression, intelligent event selection, and/or siganl-dependent data rates could greatly reduce the data volume.

In any case, it is recommended to record in parallel with the primary data stream at (say) 20 Hz sampling one or two additional data streams at, say, 1 Hz and 0.05 Hz rates that are obtained by digital low-pass filtration and decimation in the station processor. This will only insignificantly increase the data volume but greatly simlify data screening and the extraction of long-period signals. The low-rate data should be kept only for events.

Station processors and recording systems are being developed in various places, and at least one of them is expected to become commercially available within a year. It appears also possible to use one of the better personal computers as a VBB station processor, although task of writing a necessary software should not be underestimated.

3.2.4 Recording media A continuosly recording three-component VBB station with a sampling rate of 20 Hz will produce between 40 and 100 Mbyte of data per week (depending on the amount of data compression). These data can at present be recorded on one C-300 cartridge, or data from two weeks on a 6250 bpi computer tape. Developments in this field are however so rapid taht it is imposssible to recommend a specific solution. It is expected that the technology of optical disks will soon provide an alternative to magnetic recording. The small volume and greater durability of these media, as well as the rapid accesds to the data, would in fact suit seismological needs very well