Click for next page ( 26


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 25
4 The Need for Regional Seismic Networks The recognized problems with regional seismic networks examined in Chapter 3 raise the question of whether the new U.S. National Seismic Network (USNSN) would be a superior system for assessing the nation's earthquakes. Both approaches have advantages and disadvantages. The USNSN will open important new avenues for seismological research; by itself, however, it will be insufficient to meet the nation's seismic data needs. The seismic stations of the USNSN are planned to be of very high quality (see Chapter 5~; but the number of stations deployed will be small- only 100 to 150 USNSN stations are proposed compared to the more than 1,500 stations now deployed in existing regional networks. High station density is an unavoidable prerequisite to successful analysis in many types . . . . . 01 selsmo ogles Investigations. One of the important advantages of a dense network of stations is that many more small earthquakes can be detected and located. Because small earthquakes occur much more frequently than large ones and are associated with active tectonic structures, dense networks can define and resolve the dimensions and characteristics of these structures in a relatively short time. For instance, in California, the proposed station spacing of the USNSN will allow determining locations for most earthquakes of magnitude 3.0 or greater, or a few hundred events in an average year. The existing regional networks in California record all earthquakes above magnitude 1.5' or more than 20,000 events every year. Accurately locating large numbers of earthquakes is important for recognizing and defining active faults and understanding the seismotectonic structure of 25

OCR for page 25
26 ASSESSING THE NATION'S EARTHQUAKES actively deforming regions (see the Whittier Narrows case study below). Dense regional networks are also necessary for well-constrained determinations of earthquake focal depths and focal mechanisms for events below magnitude 5. These three types of fundamental dataearthquake locations, focal depths, and focal mechanisms which for many earthquakes are available only from dense regional networks, are essential to many types of seismic studies, including seismic zonation, characterization of source mechanics, and earthquake prediction. Examples of such studies were described in Chapter 2. With the recent advent of automated, real-time event location and event analysis using computers, regional networks have been able to locate and determine magnitudes for earthquakes within a few minutes of their occur- rence. This capability has greatly increased the usefulness of regional net- works to emergency management personnel. The possibility that some earthquakes could be foreshocks of larger main shocks has led the U.S. Geological Survey to issue several short-term earthquake warnings, based on data from dense regional networks (e.g., Heaton et al., 1989; Goltz, 1985~. Data from dense regional networks have also been crucial to aftershock studies and research aimed at understanding the rupture process of earthquakes. As was stressed in Chapter 3, a primary limitation of regional networks is the bottleneck created by conventional short-period seismometers linked by analog telemetry, a system that severely restricts seismic wave recording in terms of both frequency bandwidth and dynamic range of amplitudes. Integration with the USNSN would greatly improve the digital telemetry capabilities of regional networks and make it possible at some sites to deploy three-component sensors (two horizontal components in addition to the standard vertical component) with enhanced bandwidth and dynamic range. The special contributions envisioned as coming from such improved regional networks as part of a National Seismic System are described in detail in Heaton et al. (1989~. One program objective that generates considerable interest is short-term warning of imminent ground shaking. In great earthquakes that occur on very long faults, substantial damage is often experienced at large distances from the earthquake's initial rupture point. Because seismic waves travel slowly in relation to electromagnetic waves, it is possible to warn of immi- nent strong ground shaking from an earthquake that has already started by using electronic messages that can arrive several tens of seconds before the strong shaking. Upon receipt of these messages, local computer systems could trigger automatic safety responses and warnings. Heaton (1985) estimated that such a system could have provided the Los Angeles region more than a minute's warning before the great Fort Tejon earthquake of 1857, an event with a significant probability of recurrence in the next several decades. Upgraded dense regional networks could play an important role in post- earthquake disaster response and recovery. Rapid estimates of the areas of

OCR for page 25
THE NEED FOR REGIONAL SEISMIC NE7~WORKS 27 maximum ground shaking greatly facilitate search and rescue operations immediately following a damaging earthquake. If sensors have sufficient dynamic range to record on-scale both the strong ground shaking during a large earthquake and the weak ground motions during smaller events, data from the much more common small earthquakes can be used to estimate ground response during the large events. This capability,- also useful in earthquake engineering and land use planning, currently is severely lacking in the United States. As discussed elsewhere in this report, the scientific uses of high-quality, dense regional networks are numerous and include research on earthquake sources, attenuation of seismic waves, generation of building codes in areas subject to earthquakes, tomographic imaging of the earth's crust and deep interior, and identification and discrimination of nuclear explosions. The USNSN is an important step forward in providing data for the nation's seismological research. However, the USNSN, by itself, cannot fulfill many of the most important needs for seismic data across a broad spectrum of disciplines (e.g., public safety, public policy, critical facility siting, and basic science). An integrated National Seismic System (see Chapter 6), with dense regional networks in areas of high seismic potential, can fulfill these needs. The unpredictability of the earthquake process makes it difficult to site dense instrument arrays. Limited resources preclude instrumenting the en- tire country with stations spaced at 50 km or less, which is the density in southern California. The high seismic potential of the major California fault systems provides a ready justification for dense networks there. But in the central and eastern United States, defining the seismic hazard to which a region may be subjected is a much more subtle and difficult problem, a fact that has contributed to the inadequate support for dense regional networks there. The societal and scientific benefits that can accrue from dense regional networks are best shown by example. The following case studies document two earthquake sequences: the Whittier Narrows earthquake, which occurred within an existing regional network, and the Painesville, Ohio, earthquake, which occurred outside network coverage. CASE STUDY: THE 1987 WHITTIER NARROWS EARTHQUAKE IN THE LOS ANGELES METROPOLITAN AREA, CALIFORNIA The moderate-sized (magnitude 5.9) Whittier Narrows earthquake oc- curred in the east Los Angeles metropolitan area at 7:42 a.m. (PDT) on October 1, 1987; it caused three direct fatalities and damage exceeding $350 million in many communities in Los Angeles and Orange counties.

OCR for page 25
28 ASSESSING THE NATION'S EARTHQUAKES The earthquake occurred in a densely populated area, but the location had seismological advantages: the focus was beneath the overlap of two re- gional networks, the 200-element Southern California Seismic Network jointly operated by the California Institute of Technology (CIT) and the U.S. Geo- logical Survey (USGS), and the 24-element Los Angeles Basin Network operated by the University of Southern California (USC). Because of the excellent recordings of the earthquake afforded by the two existing regional networks, seismologists could quickly provide useful information to assist disaster response teams and emergency response officials. These data have also been the basis of detailed studies that have greatly improved understanding of the tectonics of the Los Angeles basin and of the seismic hazards facing the Los Angeles metropolitan area. The Whittier Narrows earthquake occurred at 7:42 a.m. By 8:10 a.m., information on the earthquake's location, accurate to within 2 km, and on the magnitude of the main shock was available to emergency services personnel. (The CIT-USGS system has since been upgraded so that such data are available within 7-10 minutes.) This information was then used to help coordinate search and rescue operations. By 11:00 a.m., a focal mechanism was determined by using the data from the regional networks; it showed that the earthquake had occurred on a west-striking thrust fault a subhorizontal fault along which the upper block had moved south, perpendicular to the strike of the fault. No such fault had been previously recognized in that area. Thus within a few hours after the earthquake, it was known that an earthquake with a magnitude of ~6 had occurred on a previously unrecognized thrust fault that could pose an additional earthquake hazard to the 12 million inhabitants of the Los Angeles metropolitan area (Hauksson, 1988, and Hauksson and Jones, 1989~. Data from portable arrays of seismometers installed in the epicentral region during the aftershock sequence were also used. However, because this aftershock sequence decayed particularly rapidly and the portable instruments were not installed until more than one day after the main shock, 90% of the magnitude >3.0 aftershocks occurred before data from the portable arrays were available. Thus, fundamentally important details of the Whittier Nar- rows aftershock sequence would have been irretrievable if data had not been obtained from the existing regional networks. Using these data, Hauksson and Jones (1989) were able to construct a detailed, three-dimensional picture of the faulting during the Whittier Narrows earthquake sequence. The main shock and about half of its aftershocks occurred on the west-striking thrust fault; about one-third of the aftershocks, including the largest (with a magnitude of >5.3 on October 4), define a steeply dipping north-northwest striking fault with oblique right-lateral strike- slip movement. The near-vertical aftershock fault defines the edge of the subhorizontal main shock fault and may have confined the main shock slip.

OCR for page 25
THE NEED FOR REGIONAL SEISMIC CORKS 29 This could explain why the patch of fault that slipped in the Whittier Nar- rows main shock was exceptionally small for an earthquake of this magnitude. The accommodation of the surrounding rock to the large strains produced by the main shock and largest aftershock was revealed in the small normal and thrust faults that were activated in the hanging wall above the main shock. The large quantity of data generated in the aftershock sequence was also used to develop a new, more accurate model of the seismic velocity structure of the Los Angeles basin. Of course, the discovery of a previously unknown fault in the middle of the Los Angeles metropolitan area caused immediate concern. The most important unresolved question was the extent of the fault was it limited to the Whittier Narrows area, or did it extend across the Los Angeles basin? Geologic investigations (Davis et al., 1989) had shown that the anticline that was the surface expression of the main shock fault at Whittier Narrows extended westward across the full width of the Los Angeles basin and sea- ward into Santa Monica Bay. Moreover, analysis of the shape of the anticline at Whittier Narrows and at another site near downtown Los Angeles strongly suggested that a thrust fault was buried beneath the anticline at those loca- tions. The geologic information could not resolve whether or not the thrust fault extended under the full length of the anticline, nor could it show if the fault was currently active. To answer these questions, Hauksson (1988) and Hauksson and Saldivar (1989) analyzed the data from small earthquakes recorded by the CIT-USGS and USC regional seismic networks. Because the networks had been in operation for many years prior to the earthquake, 15 years of pertinent data had already been archived. A search of these data for small earthquakes produced by thrust faulting showed that the full length of the anticline, from Whittier to Malibu, is indeed underlain by active thrust faults. These results have led to a reevaluation of the earthquake hazards facing the Los Angeles area. The new scientific findings are, in turn, being considered by local governments as they revise the seismic safety elements in their general plans. The data products of the permanent regional networks constitute an indispensable contribution to these important scientific and hazard assess- ment advances. CASE STUDY: THE 1986 PAINESVILLE EARTHQUAKE IN NORTHEASTERN OHIO The Whittier Narrows case study illustrates convincingly that high-qual- ity information can be recovered when an event is "captured" by a dense seismic network. The Painesville, Ohio, case study illustrates the opposite

OCR for page 25
30 ASSESSING THE NATION'S EARTHQUAKES situationthat important questions can remain unanswered when seismo- graphic network coverage is absent. On January 31, 1986, a modest-sized earthquake (magnitude 5.0) oc- curred in northeastern Ohio about 40 km east of Cleveland. It was felt in 11 states and in Ontario, Canada, and caused some minor damage (modified Mercalli scale intensity VI-VII) at distances up to 15 km from the epicenter. Its strike-slip mechanism implied a compressive east-northeast stress regime entirely consistent with previous events in the surrounding region. Thus, a first impression of this earthquake was that it was unremarkable, that is, very representative of the scattered, infrequent seismicity that characterizes much of the crust of the eastern United States. Two facts, however, resulted in a greatly enhanced level of interest and concern about the Painesville earthquake. First, it occurred within 17 km of the Perry Nuclear Power Plant and produced high-frequency accelerations of 0.18 g there; second, it occurred within 15 km of the three deep fluid waste disposal wells that had injected over 1 billion liters of fluid into the earth's crust at depths of 1.8 km and at pressures exceeding 100 bars above ambient levels. The subsequent detailed investigations of this earthquake by the U.S. Nuclear Regulatory Commission, the Ohio Environmental Protection Agency, the Cleveland Electric Illuminating Company, the U.S. Geological Survey, Weston Geophysical Corporation, and the Stauffer Chemical Company (operator of the two 1800-m Calhio wells) centered on the question of whether the fluid injection operations of the two deep waste disposal wells induced the Painesville earthquake. Several observations support a causal connection: (1) the Painesville earthquake was the largest event known to have occurred in the region; (2) in situ stress measurements in the Paleozoic sedimentary units overlying crystalline basement indicated the presence of high levels of deviatoric stress so that preexisting favorably oriented faults would be close to failure; (3) modeling (Nicholson et al., 1988) indicated that wellhead injection pressures of 110 bars could induce pore pressure changes from several bars up to 40 bars at 12 km from the wellbore, the actual value being quite sensitive to the confinement characteristics of the injection res- ervoir unit; and (4) pore pressure changes of this magnitude are known to have triggered earthquakes in other situations (e.g., Simpson, 1986~. Arguments favoring a natural rather than an induced origin for the Painesville event include the following: (1) the main shock's depth, although poorly constrained, places it in crystalline basement, not in the overlying Paleozoic rock where injection had occurred; (2) the main shock hypocenter was 12 km from the wellbores and approximately 3 km deeper than the injection depth; (3) northeastern Ohio had had a history of low to moderate earthquake activity prior to any injection operations; (4) the Painesville earthquake occurred 11 years after pumping had begun and did not correlate with any

OCR for page 25
THE NEED FOR REGIONAL SEISMIC NETWORKS 31 unusual pumping conditions; and (5) there was only one microaftershock, and no known prior seismicity in the crustal volume between the wellheads and the hypocenter. These basic facts and all other data pertaining to the Painesville earthquake have been exhaustively examined by many investigators (principally Nicholson et al., 1988), but the unsatisfying conclusion is that a definitive decision on whether the main shock was natural or induced cannot possibly be made, given the quality and quantity of available seismological information. Had a seismic network been in operation in the epicentral region, a more definite conclusion would likely have been reached. At the very least, two important additional pieces of evidence would have been available: tight constraints on the hypocenter's depth, and a much more sensitive test of whether microearthquakes had occurred around the wellbores during the 11- year pumping history. ~ focal depth confidently constrained to the S-km centroid depth estimate, for example, would put the hypocenter about 3 km deep in crystalline basement. Without appeal to special fracture or joint pathways or pore fluidespecially if an absence of microearthquakes could be confidently established for the intervening crustal volume a causal connection between the waste disposal wells and the main shock could be ruled out. An in-place seismic network could have provided the necessary information. Thus the Painesville event, by virtue of the lack of key data essential to the resolution of an important question, provides an excellent example of the value of operating seismic networks.