Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
8 DATA HANDLING, PROCESSING, AND USER SERVICES The Panel expects that during the next five years digital seismic data will become available from a variety of sources, previously discussed in detail. We urge that data from Seismic Research Observatories (SRO's), the High- Gain Long-Period (HGLP) network, stations of International Deployment of Accelerometers (IDA) program, selected data from seismic arrays, and geophysical observatories in- stalled by NASA be considered together as a single compre- hensive data baseâbut again, not to the exclusion of analog data. In addition, a number of digital arrays and single sta- tions are operated around the world by other countries, including Canada, Mexico, France, Germany, and Sweden; arrays are operated by the United Kingdom in Australia, India, and Brazil. Data from certain of these ^ources, if obtainable on an exchange basis, would complement the global data of U.S. digital stations. Efforts should be made to ascertain the availability of these data and to examine their utility to the U.S. data base. There is an immediate need to establish a centralized distribution facility for a strainmeter records, but, be- cause of the great variety of instrumentation currently employed, this may be a difficult task. However, we en- courage examination of available strainmeter data sources and hope that in the near future at least some of this type of data may be included in the U.S. data collection. Collection, data handling and processing, organization, and archiving of analog seismic data are carried out by the Environmental Data Service (EDS) of the National Oceanic and Atmospheric Administration (NOAA) at its Data Center in Boulder, Colorado. Seismograms are mailed to EDS, mostly via the USGS in Golden, Colorado, to be microfilmed and archived. The Panel encourages the work of EDS and urges 49
50 its continued funding at a level that will prevent back- logging of seismograms awaiting reproduction and the resulting delays in data orders. The potential volume of data from digital seismic stations is great, and EDS should assume responsibility for the storage and distribution of these data. It makes a great deal of sense for the same organization to handle both analog and digital data, and EDS is highly qualified for this task. Some problems in connection with the processing of digital data require examination. These include data review, quality control, and reformatting. Quality con- trol is necessary to correct errors in header information stored on. the magnetic tapes and to remove other possible sources of error caused by poor tape quality of other problems. Reformatting is necessary because most indi- vidual stations record for up to two weeks on a single reel of tape. The two-week record must be divided before the station data can be merged with that of the rest of the network. At present, data from the HGLP, SRO, and ASRO networks are first processed by the Albuquerque Seis- mological Laboratory (ASL), operated by the U.S. Geological Survey, and then forwarded to the Seismic Data Analysis Center (SDAC) for distribution. The Panel recommends that the ASL continue to carry out the basic quality control of incoming digital seismic data. If SDAC is unable to con- tinue the long-term distribution of digital data, then ASL should assume the responsibility for merging all daily sta- tion data onto a single file or so-called network-day tape. DATA HANDLING AND PROCESSING There are two basic ways to store digital seismic data, and the choice between them depends largely on the methods by which users may access the archives. The simplest method of storage is to retain the network- day tapes. Data from the HGLP, SRO, and ASRO networks are expected to generate between one and two standard 's-inch reels of tape per network day. Addition of other seismic data may increase this to three or four reels per day, totaling about l000 tapes per year from the integrated network. This is not an excessive number of tapes, and they could be stored relatively easily. The use of a computer-mass-storage device would provide a much more sophisticated and flexible system. Such a device, the Datacomputer, is now being used by the Computer
5l Corporation of America, under the sponsorship of the Defense Advanced Research Projects Agency. Storage is on high-speed video-tape devices with on-line capacity between l011 and l012 bits and virtually unlimited off-line capac- ity. The advantage of this system is that access to complex file structures is included in the softwave support system and data can be retrieved according to the require- ments of the user. In particular, the system will be able to generate an event-associated series of waveforms, as well as bulletin information. On the other hand, in a system that relies heavily on bulletin production, ex- tensive processing is necessary in order to make the proper associations. From the point of view of archiving, both systems are equally valid. Both suffer the problems of magnetic-tape storage, and the problem of tape life, in particular, should be examined carefully, whichever system is adopted. The distribution of analog data is carried out by EDS. Users request by mail copies of the data contained in the archives, usually in microfilm form, and the copies are supplied at a nominal cost. The Panel recommends long- term funding to permit the continuation of this service. Three ways a user can obtain access to the digital-data collection are described below. l. Computer Storage and Retrieval. If all seismic data are stored on a computer-mass-storage device in an event-associated form, access to the mass store will be by computer network (such as the ARPANET). Users with computers directly linked into this network would be able to retrieve directly (in a read-only mode) from the ar- chives. Other users could use standard telephone connections to obtain a slow-rate data access. One user with a direct connection to the mass store would be EDS, which would make up data tapes according to user requests and mail copies to users for a nominal charge. 2. Network-Day Storage with User Services. In this mode of operation, all steps in the processing and dis- tribution would be handled by EDS. Archival storage would consist primarily of network-day tapes. EDS would have enough computational facilities and staff to respond to a variety of user requests. For example, waveforms for a specific event could be selected from the archives and made into a tape in a format compatible with the users' processing facilities. It is anticipated that event- associated tapes could be made up on a regular basis for particularly interesting or large events.
52 3. Network-Day Storage Only. In the simplest methods, EDS would store network-day tapes and duplicate them on request. Selection between these alternatives must be based on funding. Option l is clearly the most efficient for all users, although it is also the most expensive. Option 2 would be adequate for many purposes, although the time between user request and data receipt might be extensive at times. Option 3 is, in our opinion, generally in- adequate . The costs of options l and 2 should be examined in detail, along with the advantages and disadvantages of each. A summary of current network data handling, described in part above, is presented schematically in Figure l6. THE NEED FOR REGIONAL RESEARCH CENTERS Developing competence in the use of digital data within the user community is of prime importance. Conventional NSF support, based on good research proposals, is needed for analysis of digital data. There will be a need to inform and to create interest among potential users of digital data; this can be done at symposia and in other ways. The most promising areas of digital seismic research require the development of certain basic software. One way to meet this problem would be to take advantage of the experience already gained at several institutions across the country by encouraging other workers to visit these institutions and learn first hand what can be done with digital data. The Panel recommends that several geographically dis- tributed Regional Research Centers be established. These Centers would have the following characteristics: (a) Each would have a library of digital seismic data. (b) Each would have access to extensive computing facilities. (c) They would develop a variety of software for the manipulation, display, and processing of digital seismic data. (d) Their staffs would include programmers, analysts, and research seismologists (either at the institution or in the immediate area).
53 s 5 I 0) c o +J <o 0) § h M-l I M I O C O O M ,H <1) l4-t 4J <U QJ Pn 2 g ii-1 ^ â¢H -H Q CT* D W o ⢠H D
54 (e) They would encourage visitors to spend periods ranging from a few weeks to a few months working with their staffs on research problems. Institutions with these characteristics already exist. Examples include the following: on the East Coast, the Lincoln Laboratory of the Massachusetts Institute of Tech- nology, the Lamont-Doherty Geological Observatory, and the Seismic Data Analysis Center; on the West Coast, the Uni- versity of California at San Diego and at Berkeley and the California Institute of Technology; and, in the Boulder, Colorado, area, EDS together with nearby universities. The Panel suggests that several of these institutions be des- ignated as Regional Research Centers and that interagency support be provided to maintain their capability to receive and help visiting scientists. This approach could elimin- ate or reduce the unnecessary duplication of hardware and software facilities that might accompany widespread use of digital seismic data. USER SERVICES Routine services such as the Preliminary Determination of Epicenters (PDE) and the International Seismological Centre (ISC) bulletins currently use seismological techniques that are little changed since the l930's except for the adoption of digital computers to process large amount of data rapidly. The impact of the new digital-data base on the science and on its capacity to serve the community will be immense. For example, an analyst might use a display device to cata- logue waveforms for event parameters. Such operations as spectral analysis of body- and surface-wave phases, narrow- band and matched filtering, rotation and combination of components, comparison of events in an aftershock sequence, and time-frequency analysis of short-period signals can be performed quickly and easily. It is also possible to auto- mate the timing of first arrivals and the identification of secondary arrivals. Once digital data are at hand, the techniques are computationally simple to estimate routinely the depths of the hypocenters of most earthquakes to within a few kilometers, and with that information to refine the hypocentral parameters further. We now have the opportunity to estimate routinely not only where stress is released in the earth but also how it is released. It is desirable that the possibility of routine determination of certain source parameters, such as
55 moment tensor, be considered. Existing analytical methods can be used to estimate such parameters as direction of rupture propagation, fault length, and seismic moment. Such analysis probably could be routinely applied now to all events of MS > 6 and perhaps as low as 5.5. Even if pres- ent models are wrong, we will have the parameters with which to reproduce the seismograms and thus to establish new models. The production of a first-rate earthquake catalogue is essential to the seismological community, not only for statistical studies but as a means for even-associated archiving of waveform data, including digital data. The best input data for the production of such a catalogue would be that from a primary network of accurately cali- brated, competently operated, digital-recording stations with good geographical distribution. The availability of digital data in a standard format carries the promise that a certain amount of information can be processed routinely and included in the bulletins of seismological services such as the ISC and the National Earthquake Information Service of the USGS. Further refinements in seismological services should come from intensive research, as more data become available. Such research should be undertaken as soon as possible. A. The Panel recommends continuation and improvement of user services such as the PDE program and the ISC Bulletin and the reporting of additional event parameters. B. The Panel recommends initiation of a program of intensive research to determine what impact the new digital- data base will have on user services and in what ways the new services can be implemented.
