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.
CURRENT CAPABILITY FOR EARTHQUAKE PREDICTION IN THE UNITED STATES This section discusses the present U.S. capability for earthquake pre- diction in terms of recent developments, observational capability, and laboratory and theoretical studies, including the fracture process and the state of development of physical models to explain the precursory observations. Also discussed are critical weaknesses of the program. HISTORY OF THE U.S. PREDICTION EFFORT During the early l960's, a large percentage of the seismological re- search in the United States (exclusive of that related to seismic prospecting for petroleum) was part of the Nuclear Test Detection Pro- gram of the Advanced Research Projects Agency of the Department of Defense. The Nuclear Test Detection Program in many ways converted seismology into a modern observational science, using arrays of seismic detectors, digital processing of data, and the application of informa- tion theory in signal processing. In its early stages, the Test Detection Program concentrated mainly on radical improvement of basic research in seismology and on the modernization of seismic instruments on a worldwide scale. In the early l960's, the Worldwide Network of Standard Seismographs, consisting of about l25 stations recording short- and long-period information, was installed. The worldwide nature of the network, the availability of microfilm records, and the sensitivity of the instruments used revolutionized the experimental side of seismologi- cal research. In l965, following the great Alaskan earthquake of March 27, l964, an ad hoc panel of eminent scientists, appointed at the request of the President of the United States, drew up a l0-year plan for a major U.S. program in earthquake prediction. Both the U.S. Geological Survey and the U.S. Coast and Geodetic Survey began small-scale efforts in research on the subject in the following year, and interest rose among university scientists, who were supported mainly by the National Science Foundation. No major federal effort was mounted for earthquake prediction, however, until l973, when funds for this purpose were added to the U.S. Geological Survey's budget and the seismological research effort of the National Oceanic and Atmospheric Administration of the Department of Commerce was transferred to the Geological Survey in the Department of the l5
l6 Interior. By l973, because of these efforts and because of the promis- ing results of the observations by scientists in Japan and the USSR, seismological researchers in the U.S. began to turn their attention to earthquake prediction. The U.S.-USSR Working Groups in Earthquake Prediction have established a relatively small but quite vigorous program of joint field, labora- tory, analytical and theoretical studies in earthquake prediction and related fields. This program was established under the aegis of the l972 Agreement between the U.S. and USSR on Cooperation in the Field of Environmental Protection. The work has been organized into four areas: (l) field investigations of earthquake prediction; (2) labora- tory and theoretical investigations of the earthquake source; (3) mathematical and computational prediction of places where large earth- quakes occur and evaluation of seismic risk; and (4) engineering- seismological investigations. Successful projects to date include the establishment of a joint seismograph network near Garm, Tadzhik SSR, to search for seismic forerunners to larger earthquakes; the establishment of a joint seismograph network around the Nurek Reservoir, Tadzhik SSR, to study reservoir-induced seismicity; the study of the spectral content of earthquakes using the Soviet "frequency-selecting" seismograph system and American broad-band, digitally recording equipment; collaborative laboratory studies of the effects precursory to failure and sliding in rock samples; the application of pattern-recognition techniques to the prediction of earthquakes; the establishment of a joint strong-motion network in the Tadzhik SSR; and the collaborative instrumentation and testing of buildings subjected to earthquake-like motions simulated by explosions. Exchange of information with Japan is facilitated by joint Japanese- U.S. symposia on earthquake prediction, supported by the National Science Foundation. Four meetings have been held, in Japan (l964) and in the U.S. (l966, l968, and l973), under the auspices of the U.S.-Japan Cooperative Science Program. In l973, the RANN Division of the National Science Foundation became a major source of support for engineering seismology and for research into public policy issues related to earthquake prediction. The U.S. Nuclear Regulatory Commission is now supporting seismological research concerned with seismic hazards to nuclear power plants and the seismo- logical criteria for siting nuclear power plants. The National Aeronautics and Space Administration has started a long-range program aimed at adapting technology from the space program to possible future uses for earthquake prediction and hazard reduction. The responsibility for issuing earthquake predictions and other assis- tance measures specified in the Disaster Relief Act of 1974 was assigned to the U.S. Geological Survey in l975. OBSERVATIONAL CAPABILITY The success of earthquake prediction depends on appropriate field ob- servations more than on any other factor.
l7 Instrumentation Most instrumentation likely to be useful in field monitoring has been developed or is in an advanced stage of development. Improvements in accuracy and stability, fieldworthiness, and installation techniques are needed for certain devices. In particular, strainmeters, geochemi- cal monitoring instrumentation, and 3 wave-length electro-optical ranging devices, among others, still require some development. New instruments that yield great improvement in resolution have also been proposed and may ultimately prove useful, especially if they can be installed at adequate depth to remove environmental noise effects. At present, however, most development has been completed or is near com- pletion, and rapid deployment of a wide variety of sensors is possible. Considerable improvement in resolution still seems likely in geodetic systems based on satellite and extra-terrestrial positioning systems. Seismic Stations Most of the field-monitoring effort at present is concentrated in California. About 300 seismic stations have been installed, and about half of them are involved in experiments related directly to prediction of earthquakes. The remainder can be used indirectly for earthquake prediction since their function is to locate small earthquakes. The time distributions of earthquakes, e.g., foreshocks, may be useful in earthquake prediction. Improvement in evaluation of real-time seismic data from these networks is needed. About l20 other instruments have been installed in California for the continuous monitoring of strain, tilt, and fault creep, many in the past year. Because the instruments are too sparsely distributed, no clear precursory signals have been detected by more than one or two instru- ments for any earthquake. The present networks can be expected to re- cord an earthquake of magnitude 5 or greater, in a densely instrumented area, every 3 to 4 years. A smaller number of similar instruments have been installed in Alaska, Nevada, Utah, Missouri, Washington, and New York. Crustal-Strain and Elevation Measurements Some l,200 monumented lines, 20 km in average length, have been mea- sured by laser-ranging devices to accuracies of 5 x l0- 7 strain. Yearly, 500 such lines are measured, in California and Nevada primarily, and several thousand kilometers of first-order level lines are available for re-leveling in seismically active areas of California and Nevada. The repeat intervals are long, greater than 5 years for most lines (though more frequent for a few). The leveling data have recently proved to be the basis for finding a large area of possibly precursory uplift in Southern California along the San Andreas Fault. Very little leveling, 300 km per year, is done specifically for earthquake research, however, so that the bulk of the measurements are not made on a timely schedule.
l8 Numerous instruments for the local measurement of strain are opera- tional, including such devices as laser strainmeters. Other Field Measurements Measurements are made intermittently along a 40-km-long section to de- termine the electrical resistivity of the San Andreas Fault in Central California by active and passive methods. Self-potential measurements are being conducted at about 20 sites in central California. Radon emanation from soils and subsurface waters is monitored weekly at about 30 sites in California, and measurements also are carried out at Blue Mountain Lake, in New York State. The level of effort is small and evaluation of the technique will require a decade or more at pres- ent levels. An array of 7 magnetometers with l/4-gamma sensitivity is operating in a continuously recording differential mode in the densely instru- mented section of the San Andreas fault near Hollister. One of the best-defined precursory anomalies yet observed in California was re- corded on the San Juan Bautista magnetometer prior to the l974 Thanks- giving Day earthquake (m = 5.2). No other anomaly of comparable duration and signal-to-noise ratio has been observed in the 2 years of recording on any of the magnetometers in the array. Surveying with magnetometers by registering differences at sites spaced l0 km apart has been conducted semi-annually along two long lines in California. This is an inexpensive technique for searching for long- term changes in a local field. Gravimeter surveys designed to detect elevation changes of greater than a few centimeters are possible now, but have only been attempted on a limited basis. Use of ground-water-level variations for predicting earthquakes has received little attention here by comparison with efforts in China. A few wells are now being monitored in the Hollister area. Aberrant animal behavior has been noted, especially by Chinese ob- servers, but no systematic program to search for such possible precur- sors has been initiated in this country. LABORATORY STUDIES Most of the conceptual underpinnings of the physical models now used to explain earthquake precursors derive from laboratory experiments. Dilatancy and precursory fault creep, both instabilities that lead to failure, were observed in the laboratory long ago. The search for other possible precursors, e.g., changes in electrical resistivity, seismic velocity, and microseismicity, was begun by workers in the U.S. and Japan a decade ago under the controlled conditions available to labora- tory experimentalists. Field experiments designed to test some of these models for the pre-failure process have lagged because they are difficult to perform. On the other hand, precursory phenomena observed in the field have guided laboratory experimentalists in recent years to
l9 rapid advances in our understanding of the physical bases of earthquake phenomena. Roughly 30 percent of the present laboratory capability for high- temperature and high-pressure rock-deformation experiments is used for relevant earthquake-prediction research. Two of these laboratories have been very productive training centers for geophysicists currently active in earthquake research. Experiments on the rheological (flow) behavior of rocks have pri- marily been concerned with steady-state flow at high-temperatures. The rheology at moderate temperatures is experimentally more difficult, and is currently given little attention. The subject is an important one, however, because deformation of the lower crust and uppermost mantle is involved in large earthquakes. Most current research focuses on the details of the failure process preceding brittle fracture, which is, in effect, the "laboratory earthquake." Dilatant cracking and fault creep precede the sudden failure, and accelerate unstably very near the time of fracture. New experiments are under way to study these processes at greater than room temperature. Studies of faulting as a dislocation along a sliding surface between two large blocks are now under way and should yield direct observation of precursory phenomena and earthquake source parameters. MODELING OF EARTHQUAKE PHENOMENA Part of any large-scale program of earthquake prediction must involve the development of models of the earthquake process. These conceptual models are syntheses of laboratory and field observations and interpre- tations. They are of necessity simplifications of the complexities of nature: irregular geometries are modeled by simple ones; empirical designations of rheological processes are fitted by simple functional expressions. Evaluation of the consequences of predictions from theo- retical models has a feedback on the laboratory-experimental and field- observational programs. One recent example of this feedback has been an attempt to assess whether it is possible to determine conventional focal parameters of earthquakes, such as stress drop, magnitude, fault length, etc., from seismograms, and indeed what parameters should be used to describe an earthquake. Models of the pre-history and history of a seismic event must have three ingredients: (l) the rheological relationships connecting the stresses in the rocks and the deformational response of the rocks to those stresses; (2) the distribution of the sources of force (stress) that ultimately provide the impetus for a repetitious, sequential earth- quake history; and (3) the geometrical constraints on a dynamical earth- quake system. Intimately connected with the need to know this information for the relevant parts of the earth is a need to have reliable information regarding certain basic geophysical parameters: temperatures in the earth's interior, perhaps to considerable depth; the location and characterization (size and physical properties) of major inhomogeneities in the crust and upper mantle of the earth; and
20 as much characterization of the properties of materials (such as porosi- ties, elastic parameters, etc.) in seismically active regions as possible. Temperatures are important because the rheological proper- ties of matter are strongly temperature dependent, as well as dependent on many other parameters, including stress and strain rate. Characteri- zation of lateral inhomogeneities is important especially in seismic regions near ocean-continent boundaries, such as California and Alaska, where the thicknesses and physical properties of various major layers in the earth's interior are changing. Porosities influence the role of water in the entire sequence of events leading up to and following an earthquake. Rheological Information The rheological relationships most often used in building models come from laboratory investigations. Unfortunately, these experiments are difficult to carry out under conditions that closely simulate the actual earthquake environment. These laboratory results are obtained from ex- periments that are necessarily brief (minutes, hours, or days) compared with the recurrence rates of large earthquakes (tens and hundreds of years). Hence, in this and in other areas of the prediction problem, there is considerable extrapolation based on the time-scale differences. An assumption often is made that the relationship of magnitude to pre- cursor time for large earthquakes can be derived from that for small earthquakes. The immediate vicinity of shallow earthquake faults is described by the rheology of brittle fracture, especially in the stages before earthquakes when the stress fields are relatively large and the temperatures relatively low. But in devising a systematic approach to modeling, a rheology of the earth relatively far from the active seg- ments of earthquake faults, where the stresses may be relatively lower, is also important. Parts of the crust and upper mantle must also move in large earthquakes, albeit more slowly. Physical models of the re- sponse of materials to low stresses, in both high- and low-temperature regimes, will help determine the amount of intraplate strain accumula- tion as well as the motions of the earth in the region vertically below active zones of shallow earthquake faulting, as in California. Are the present rheological models, often obtained from steady-stress laboratory experiments, applicable to the transient physical changes that charac- terize earthquakes, and to the associated slower creep motions of the regions far from the active fault zones? What is the response to applied stresses of a system permeated by many flaws (e.g., dilatant cracks)? What is the behavior of earth materials approaching ultimate failure? Is there a different rheological behavior for unfaulted material than for a material crossed by major faults? We cannot yet answer these questions in detail, and models of earthquake events usually assume simplified answers in order to be able to proceed with the analysis.
2l Source of Deformation Most models of the primitive sources of power for earthquakes refer to plate tectonics to describe the mobility of the outer parts of the earth. Dynamical descriptions of the means of propulsion of the plates often refer to the presence of heat sources; these heat sources, and boosters to the driving system such as phase transformations, are usually imagined as applying stresses to a relatively rigid lithosphere (the outer l00 km of the Earth). But is the lithosphere completely rigid, except in the neighborhood of earthquake fault zones, or is it capable of absorbing strains? Do intraplate earthquakes indicate the presence of inhomogeneous stress fields at distances of a few hundred kilometers or more from major fault zones, or can these events be ig- nored in the modeling of major fault zones such as the San Andreas? Information about the relative motions within and between plates can be obtained by long-range geodetic studies, such as very-long-base-line interferometry, multilateration techniques, etc. The role of deforma- tions derived from plate-tectonic sources in contemporary models is to provide a build-up in the neighborhood of an earthquake fault of stress that is uniform with time. But is the assumption of a uniform rate of increase justifiable? Can rates of motion of plates derived as averages over millions of years of Earth history, be applied to obtain rates of recurrence of large earthquakes? Investigations of the distributions of stresses capable of causing individual earthquakes must also focus on the residual stresses left behind in the wake of earlier earthquakes. The stress field after an earthquake represents a prestress of an earthquake fault for subsequent events. But these residual stresses must also undergo relaxation due to aftershocks and aseismic creep. The degree to which faults are in- homogeneously prestressed by residual stresses, and indeed to which the regions within some tens of kilometers of the fault (and by extension the entire intraplate space) are inhomogeneously stressed, may be im- portant in the modeling problem, since it is most likely that these residual stresses considerably affect the conditions of occurrence of future events. Geometrical Influences The orientation of earthquake faults, offsets of these faults, the way in which they terminate, and the distribution of inhomogeneities in physical properties (including rheological properties) are all complexi- ties that influence the construction of models. In most cases, these factors are unknown in the real earthquake environment because of the presence of geological complexity, including burial of earthquake faults. Mathematical and Numerical Models Thus far, theoretical computations have been for extremely limited and simplified systems; even these are very complicated to evaluate.
22 Numerical models of simplified systems have been computed at great expense, but the ability to generalize from these is questionable. Results from these theoretical studies have thus far had little impact on broad-scale programs of observation and laboratory work except to indicate the direction such work might take. SOME PROBLEMS AND DEFICIENCIES Over the past few years, important first steps have been taken and a pilot program has been launched in the United States for earthquake prediction. As significant as this effort is, it is much too small in comparison with the magnitude of the problem, and it has many critical omissions and serious weaknesses. The greatest weakness in the present U.S. program of earthquake pre- diction is the inadequacy of field projects aimed specifically at de- tecting and understanding earthquake precursors. The U.S. effort is limited both in the kinds of observations and experiments performed and in areal coverage. Field observation is concentrated in California, leaving other seismically active parts of the country essentially un- covered. However, even in California, because of limited resources, not all necessary field measurements are being made. The People's Republic of China, the USSR, and Japan are several years ahead of the U.S. in the field observations aimed specifically at de- tecting and understanding earthquake precursors. Most of the data on changes in electrical resistivity, radon emanation, changes in water level in wells, and changes in land elevation come from these three countries. As another example, basic research during the last l5 years on flow of fluids in porous media, crack propagation, dilatancy, and physical properties of rocks has played a pivotal role in development of current theories of the physical basis for precursory effects of earthquakes, but much is still unknown. It is clear that a great deal of additional basic research in these areas will be needed before it will be possible to predict earthquakes on a routine basis in the United States. Theo- retical modeling is also necessary as an aid in establishing the physical basis of earthquake prediction, and there is an urgent need to improve our observations of geochemical indicators such as radon, our in-situ stress measurements, our rock-mechanics research, and our theoretical studies of the dynamics of faulting. Geodetic measurement of crustal movements is probably the most widely used technique for earthquake prediction in China, the USSR, and Japan. Geodetic surveying is very time-consuming and expensive, however, and comparatively little effort has been devoted to such measurements in the United States. Attempts have been made in the United States to develop other instruments, such as tiltmeters, in the hope that they might, at less espense, detect the same kinds of crustal movements. But it is not certain that these other methods will be as successful as geodesy. The development of technology for measuring elevations rapidly and inexpensively with an accuracy ranging from millimeters to a few
23 centimeters could essentially replace the time-consuming process of geodetic leveling. With these accuracies, measurements made only weeks or months apart could show small but significant changes in elevation. Obviously, there is a great need for state-of-the-art geodetic measure- ments with earthquake prediction specifically in mind. At present, very few measurements of strain and tilt over baselines ranging from meters to kilometers in length are being made in the United States with earthquake prediction specifically in mind. The Japanese, on the other hand, have installed long-base-line tiltmeters at l7 stations in various parts of the country. Many Japanese scien- tists feel that very short-base-line observations of tilt (for lengths of less than one meter) are likely to sense local inhomogeneities rather than precursory effects of earthquakes. New generations of inexpensive strain meters and long-base-line water-tube tiltmeters are now becoming available. It is extremely important that such instru- ments be installed in many of the seismic areas of the United States, including, for example, the area of current anomalous uplift near Palmdale, California. In particular, it is important to establish several small arrays, with dimensions of about one kilometer, that would include several types of strain- and tilt-measuring instruments, as well as to perform repeated geodetic leveling of the array. Such arrays, which might also include gravimeters of micro-gal accuracy, would help to answer unresolved problems about measuring tilt over very short baselines and to ascertain which of the various techniques are most reliable for earthquake prediction. Although these deficiencies are serious, they could be corrected in a relatively short time with the proper emphasis and increased level of effort. Trained personnel, laboratories, and instrumentation exist. Any needed improvements in these areas can be made in a few years.
