113
STRONG MOTION SEISMOLOGY
D. E. Hudson*
ABSTRACT
Strong motion accelerographs and data processing systems for the measurement of strong ground motion of damaging earthquakes are described with some comments on future developments. Some information is given on the earthquakes and instrumentation sites for which important records have been obtained. Studies of the areal distribution of ground motions during earthquakes are discussed in terms of transmission path effects. Some speculations are advanced concerning the establishment of limiting values of earthquake ground motions.
INTRODUCTION have been based on such data, and (4) To suggest some lines of development for In the pre-instrumental era of earthquake future investigations. engineering, much was learned about the effects of earthquakes upon structures by STRONG MOTION INSTRUMENTATION a careful study of the damage caused by large earthquakes, and in this way many Because instrumentation to measure of the basic principles of earthquake strong ground motion is at present in a engineering became known in a qualitative rapidly developing stage with many current way. A more exact study of such problems proposals for new types of instruments and became possible only when the actual motions new approaches to network design, it will of the ground during strong earthquakes be of importance to give some background were directly measured in the early 1930's. detail. To measure the destructive ground Since that time and particularly since the motions associated with damaging earthquakes, late 1960' s, a rapid growth in the deployment the engineer requires an insensitive of special instrumentation to record strong instrument which will remain on-scale during ground motions throughout the seismic regions the largest ground motions likely to be of the world has resulted in the accumulation encountered and which will faithfully of a large data bank of useful records. record over the whole frequency range of These records have provided the fundamental structural importance. This has been information for a new science of strong motion accomplished by the development of the seismology and for the development of strong motion accelerograph, which records increasingly accurate quantitative methods accelerations up to one g or over with a in earthquake engineering. Until very resolution of the order of 0.001 g and a recently in the history of seismology the frequency range of 0.06 Hz to 25 Hz. There attention of geophysicists has been directed are now some 5,000 instruments of this kind almost entirely to far-field phenomena distributed unevenly throughout the seismic involving measurements of very small ground regions of the world - approximately 1,500 motions at large distances from earthquake in the United States, 1, 000 in Japan, and sources. Development of the required several hundred each in New Zealand, instrumentation and the analysis of data Yugoslavia, Iran, Mexico, etc. New Zealand for damaging earthquake ground motions have has contributed in a major way to the been carried out by earthquake engineers. development of such instrumentation through Recently seismologists have become more the MO 2 accelerograph, which in the 19608s interested in such subjects as earthquake was the first relatively low-cost instrument source mechanisms which also require which could be readily installed in the measurements of strong ground motions near large numbers necessary for adequate coverage. the source, so a fruitful collaboration between seismologists and earthquake engineers Rough estimates of the expenses of in the problems of strong motion seismology maintaining networks indicate that the is beginning to develop. direct cost of each useful strong motion accelerogram is of the order of $10,000 , The purposes of the present survey are which suggests that a considerable data (1) To briefly describe the instruments processing effort is justified in the interest required for such investigations and the of acquiring the maximum amount of information networks and data processing facilities that from each record. The data processing system are presently available, (2) To summarize should be thought of as an integral part of the characteristics of the recorded earth• the whole accelerograph network and is in quakes and of the instrumentation sites and fact the element for which the most significant to evaluate the adequacy of the data base, improvements in information collection are (3) To outline some significant studies which likely to be realised. The most important advance in data processing made in recent * California Institute of Technology, Pasadena, years is the routine introduction of CA 91125; Visiting Erskine Fellow, Depart• digital filtering. This filtering technique
ment of Civil Engineering f University of based on running weighted averages is Canterbury, Christchurch. particularly well suited to digital
BULLETIN OF THE NEW ZEALAND NATIONAL SOCIETY FOR EARTHQUAKE ENGINEERING, VOL.10, NO. 3, SEPTEMBER 1977 114 computations and has made it possible to carry Recent advances in digital instrument• out accelerogram integrations and transducer ation and in integrated circuit technology corrections over a considerably wider would seem to make it feasible to produce frequency range than had been possible by a practicable digital field accelerograph, other techniques. Not only can new records although the cost of current prototypes be routinely processed to a higher order having essentially the same overall of accuracy, but it is feasible to reprocess specifications of accuracy, dynamic range some of the old photographic paper records and frequency response is about twice that dating back to the 1930's and to recover of the standard analog photographic acceler• considerably more information from them ograph . An additional disadvantage of than had originally been thought possible. currently available digital accelerographs is the somewhat larger standby power The existing field accelerographs requirements which reduces the non-external produce records in the form of an analog power operating life by a significant amount. photographic trace on a 3 5mm or 70mm film. An advantage of the digital system is the For detailed analysis of the information, possibility of providing a short memory to the analog record is digitized using a recover the earliest portion of the triggering semi-automatic machine which combines hand- ground motion. In view of the relatively eye setting of a cross hair on the analog quick action of modern vertical triggers, trace with automatic readout and recording the memory is probably justified only if of time and acceleration coordinates on the additional complexity and power require• punched cards or magnetic tape. Currently ment does not compromise the field reliability under development is a completely automatic of the accelerograph. Digital accelerographs type scanning digitizer which will require would also be somewhat more dependent on operator intervention only for those portions laboratory-based playback equipment, since of the record which involve ambiguities or it will usually be essential for preliminary defects. A critical advantage of the current inspection and analysis to have an analog optical-photographic analog recorder plus record. It is also likely that more automatic digitization is that it permits elaborate field test equipment will be the widely dispersed field elements to be required with a correspondingly higher of a very simple form while the more complex level of training of maintenance personnel. digitization apparatus can remain in the The long range potential of such digital laboratory. Thus the special advantages of systems, however, is so attractive that an both analog and digital systems are combined extended period of careful field testing in an optimum way for this particular and evaluation is certainly justifiable. application. The first practical applications of completely digital systems will probably be in large A considerable effort is now being special instrument arrays rather than in made to develop a digital field accelerograph the widely dispersed individual stations which would directly produce the basic record which have characterized strong motion on computer-compatible digital magnetic tape. studies in the past. This would of course eliminate the need for the relatively laborious semi-automatic The overall capabilities of any measure• digitization of the analog photographic ment system can be measured by the relation• traces, at a cost of a more complex ship between noise characteristics which installation in the.field. The desirability limit the ability of the system to produce of such digital field systems depends- upon low level analyzable records and the size the extent to which rapid digitization of of the signal to be measured. As an example records is believed to be important. In my of the current state of the art, the perform• opinion digitization requirements do not ance of the instrumentation and data process• represent a decisive disadvantage for the ing system used in the present United States film recording system for the following strong motion accelerograph network is reasons. First, a large amount of information summarized in Fig. 1. Since frequency of immediate practical importance is quickly analysis of records is a frequent end available on the analog record without product, the instrument signal level can digitization or further processing. be conveniently represented by the Fourier Practical decisions which must be made promptly amplitude spectra which can be directly after an earthquake, such as the need to compared with the Fourier spectra of the evacuate a damaged structure or to empry a noise in the signal. For practical purposes reservoir, can be made from the analog this is equivalent to a comparison of the record and are not likely to be significantly earthquake undamped velocity response modified by additional data processing. In spectra with the response spectra of the fact, the analog film record has many background noise. From the left-hand portion advantages over the digital tape for this of Fig. 1 it will be seen that with the purpose since it can be developed with present system earthquakes of magnitude much faci1ities readily available almost anywhere less than M = 4 will involve ground motions in the world and the significant features even in the epicentral regions so small can be quickly determined by persons with that they will not be distinguishable on a minimum of special training or experience. the record from noise. From the right-hand Secondly, even though a large earthquake portion of Fig. 1 it will be seen that for should occur near a dense accelerograph an earthquake of given size, in this case network it is not likely that there will be M = 6.5, the ground motions will drop more than a few dozen key records requiring below measurable size at distances of the immediate attention. For the San Fernando, order of 100 km to 200 km. In this way California, earthquake, for example, although the capability of the system to adequately there were a total of 241 records almost all sample a given area can be ascertained and of which proved to be of ultimate interest, an optimum network design can be accomplished. there were only a half dozen or so near the epicenter of such special interest that THE STRONG MOTION ACCELEROGRAM DATA BANK rapid processing was of importance. As an example of the basic data 115
available for the study of strong ground of visual impression a 1 so emphasizes some motion, the contents of the Caltech data of the problems of defining a typical bank of uniformly processed records will earthquake. be briefly described. This data set com• prises 3 81 three-component accelerograms, It will be noted that the largest 187 of which are ground sites, the rest earthquake in Fig. 2 is the M = 7.1 being from upper locations on structures. Olympia , Washington , earthquake. The The ground site records are usually from the largest earthquake for which any strong basements of buildings, and it has been motion record exists is the M = 7.7 Kern ascer _ained that over most of the frequency County, California, event, recorded at a range of structural interest these basement distance of 60 km. We have no accelerograms records with a few exceptions are good, for such great earthquakes as the 1906 San estimates of free-field ground motion. Francisco, the 1923 Tokyo, the 1960 Chile, The site conditions can be roughly classified the 1964 Alaska, or the recent devastating as about 60 percent soft-alluvium, 10 per• major shocks in China. This is one reason cent hard-rock, and 30 percent intermediate. why a good deal of theoretical work is In the present state of limited knowledge needed to extrapolate available records as to the details of site conditions, this to other situations. A hopeful factor is is about as fine a classification as is the present belief that great earthquakes feasible. An important task for the future involve very large generating areas, so is to collect more detailed information on that much of the energy release must come site conditions and local geology for from points far away from a given point. instrument sites for which important acceler• A great earthquake in fact appears to be ograms have been obtained. The total number a sequence in time and space of the kinds of earthquakes represented in the data set of smaller events seen in Fig. 2. If this is 57, ranging in size from M = 3 to a maximum is the case, the maximum amplitudes of ground of M = 7.7. The maximum number of records motion at a particular point may not be at one site is 16 at El Centro, California; bigger for a great earthquake than for a the maximum number of records from one smaller shock, but the duration of the earthquake is 241 for the 1971 San Fernando ground motion would be longer. This points earthquake, of which 98 are ground motion up the importance of duration for engineering records. studies and means that additional invest• igations are required of the way structures The data bank described above is supple• deteriorate during extended applications mented in a very important way by a number of cycles of alternating loads. of more recent earthquakes, plus several dozen strong motion records from other It would be expected that a study of countries which have been obtained with such near field records would throw some similar instrumentation. There is also a light on the important problem of establish• large collection of records from the New ing upper bounds for earthquake ground Zealand and Japanese networks, but only a motion. It is usually supposed that the few of these accelerograms are of damaging strength properties of the earth's crust ground motion. In most cases the recording impose some limits to the amplitude of sites have either been distant from the ground shaking, but ideas as to what these earthquake or the earthquake has been a limits might be are being gradually scaled relatively small one. upwards as more records become available. For some years the 1940 El Centro earthquake NEAR FIELD ACCELEROGRAMS was the most damaging earthquake ground motion that had been measured, and for a Of the more than 1,000 useful accelero• time the acceleration level of about one-third grams obtained since the first record from g was thought to be near the upper limit, the Long Beach, California, earthquake of at least for the geologic conditions present 1933 , most have been at considerable distances there. Then the 1966 Parkfield earthquake from relatively small earthquakes. Only a came along with a fifty percent g acceleration few dozen have been recorded at distances and did very little damage. In 19 67 the from points of large seismic energy release Koyna earthquake in India with a two-thirds as small as the source dimensions of the g acceleration for a time held the record. earthquake. These may be called near field The present champion is of course the 1971 measurements, and they describe the most San Fernando earthquake with one and one- severe ground motions and supply the most quarter g. Since then there have been a information about earthquake mechanisms. number of small earthquakes having acceler• ation peaks as high as three-fourths g. Figure 2 shows a collection recently It is of course now well understood that made of all available near field accelerograms, the amplitudes of the high frequency giving one horizontal component of motion. acceleration peaks may depend greatly on All have been drawn to the same amplitude the characteristics of the recording and time scales. It will be noted that there instruments, and that these high frequency are striking differences in the size and peak accelerations do not necessarily bear appearance of the records, which suggests any relationship to the damage potential that the visual character of the accelerograms of the earthquake for most engineering may be an important indicator of significant structures. aspects of the earthquake which may be difficult to describe in a quantitative way It now seems likely that if sufficiently through such parameters as peak values, high frequencies are allowed there is no durations, or spectral properties. This is upper bound for ground accelerations. Perhaps another aspect of the well known problem a more fruitful approach is to examine upper that no simple set of parameters or combin• bounds for ground velocity and from them ations thereof seem to correlate very roughly estimate the acceleration levels well with the general impression of the corresponding to frequencies of structural damage potential of the event. This diversity interest. Such limiting ground velocity 116
estimates can be based on impulse momentum responsef It has so far not been possible principles applied to a simple uniform shear to explain the patterns in terms of simple fault model. In this way it can be shown factors such as distance from or orientation that maximum ground velocity should depend with respect to known faults, depth of upon the velocity of propagation of shear alluvium, soil types, etc. A typical puzzle waves, the shear modulus of elasticity, is that while at most stations all horizontal and the ultimate shear strength of the components are of the same magnitude in crustal rocks. Numerical values of these all directions, at Station SL the motions parameters are very difficult to estimate. are almost unidirectional. No explanation In particular, the effective ultimate shear of this in terms of geologic structure or strength is a very uncertain factor in view anything else has been found. of the complex roles played in the fracture process by friction, fluid pressures, and That such distribution patterns would drastic modifications of the rock material be expected to be very complex follows in the immediate vicinity of faulting or directly from a consideration of what fracture. Direct measurement of the alteration happens to seismic waves on their way from of the stress field during earthquakes had a source region to a given site. Figure 5 so far not been possible to attain, although illustrates the nature of this problem. meaningful approximations to the stress Surface motions at a particular site can drops involved can now be calculated. With be influenced in a major way by surface some reasonable guesses it appears that the topography, by subsurface irregularities, limiting ground velocity could be of the order by the presence of various wave types such of 5 m/s to 10 m/s. The highest ground as surface waves or guided waves, by wave velocity so far measured is a little over focussing, and by local geologic and soil 1 m/s. This order of magnitude agreement conditions. A common engineering approach indicates that the physical picture is is to account for variations of surface probably roughly correct, but it is still motion in terms of the local soil profile. not good enough for the engineer. Much By modelling the soil as uniform horizontal more elaborate models have been suggested, layers subjected to vertically arriving but they are difficult to justify and some• shear waves, a relatively simple mathematical times seem to be mainly arguments for believing model is achieved which has enough parameters that the highest previously measured values in it to account for almost any variation are near the upper bounds. It is clear that in surface motion. But the local soil this basic problem requires much additional environment is evidently only the last link work. in a long chain, including many other factors which could have an even greater CHARACTERISTICS OF STRONG GROUND MOTION effect on the surface motions than the local soils. The agreement of such simplified In order to study in detail the charact• models with particular situations should eristics of strong ground motions and to perhaps be regarded in some cases as separate the effects of important variables, expressing empirical relationships, and numerous records are needed at different the apparent success in some cases may sites for the same earthquake and of differ• even tend to conceal the real nature of ent earthquakes at the same site. The best the problem. To bring some of these example of the first situation is the 1971. additional factors into consideration, San Fernando earthquake with its 241 good other types of mathematical models have accelerograms, and the best example of the recently been explored. An example is second situation is the accelerograph site shown in Fig. 6 which shows that a simple at El Centro at which 16 different earthquakes geometric inhomogeneity in the presence have been recorded. Some studies based on of shear waves arriving at various angles each of these data sets will now be summarized. can account for major variations in surface motion. In Fig. 3 are plotted all of the peak ground accelerations measured during the Examining next a different type of San Fernando earthquake. Peak acceleration problem, it will be noted that the 16 is in general a parameter of dubious value earthquakes that have been recorded at the which samples only the high frequency portion El Centro site can be grouped in several of the spectrum. In this case, however, subsets, each containing several earthquakes. since all points are from the same earthquake, One set includes earthquakes occurring at some meaningful comparisons can be made. One essentially the same epicenter, and other is impressed by the great scatter of values subsets include earthquakes at various at a given distance and by the difficulty distances and azimuthal directions. This of sorting out different site conditions. makes it possible to separate the effects This overall picture is typical of other of source characteristics and transmission earthquakes for which more limited data sets paths from the local soil and geologic have been available. conditions which are of course the same for all records. Figure 7 shows some of If one examines the pattern of distrib• the results by means of a comparison of ution of ground shaking during an earthquake, the Fourier amplitude spectra of the one is usually struck by its complexity and recorded ground motions. The lower curves by the difficulty of explaining even the show spectra from four different earthquakes major features. As an example Fig. 4 shows originating near the same point and thus measured ground motions in a small area some having the same transmission path. No distance from the San Fernando earthquake. clearly defined peaks involving all or The 19 measurements in about 100 km2 is the several of the curves are apparent, and densest coverage so far achieved anywhere. this suggests that in this case the source The heavy lines have a length proportional mechanism is responsible for some of the to the response spectrum value at 0.75 significant frequency characteristics of seconds and 10 percent damping, and the the record. direction is that of the maximum horizontal 117
The upper spectra of Fig. 7 are from Strong Ground Motion", Bull. Seism. earthquakes having different transmission Soc. Amer., Vol. 65 , No. 1, February paths - here again, one notes the absence 1977. of prominent frequency peaks which could 7. Trifunac, M. D., "Preliminary Empirical be attributed to local soil conditions. Model for Scalina Fourier Amplitude Repeated attempts have been made at the Spectra of Strong Ground Motion in El Centro site to determine what might be Terms of Earthquake Magnitude, Source called a predominant site period, using to Station Distance, and Recording earthquake records, microtremor studies, Site Conditions", Bull. Seism. Soc. etc., to no avail. This is a point of Amer., Vol. 66, No. 4, August, 1976. some significance since some earthquake 8. Udwadia, F. E., Investigation of codes have proposed relationships between Earthquake and Microtremor Ground soil factors and site periods. Motions, Report EERL 72-02, Earthquake Engineering Research Laboratory, As a final type of investigation based California Institute of Technology, on ground motion measurements, a brute 1972. force statistical approach using all the recorded ground motions from all sites and all earthquakes can be attempted. Figures Paper received 17 August, 1977. 8 and 9 show the results of one such study. Clear trends are certainly present, but the wide scatter is a disturbing reminder of the smallness of the sample and the complexity of the problem. According to one point of view, such statistical studies should be pushed forward using the best available small sample theory, with the hope that eventually some meaningful probabilistic statements will emerge. Another school of thought believes that the sample is so small and the rational physical models are so complex that such statistical statements may mean less than nothing and could even be misleading„
In conclusion it may be remarked that the difficulties alluded to above in many of the examples may be discouraging but they are probably temporary. With the recent great expansion in the number of strong motion accelerographs being installed in the world, it may be expected that in the next few years an increased number of accelerograms will help to resolve these problems. Considering the cost and difficulty of installing and maintaining suitable instrumentation, earthquake engineers are making a heavy"investment in these basic strong motion seismological measurements. It is to be hoped that an equally successful effort can be applied to the interpretive problems necessary to put the results of these measurements to practical engineering use.
REFERENCES
1. Hudson, D. E., "Local Distribution of Strong Earthquake Ground Motions", Bull. Seis. Soc. Amer., Vol. 62, No. 6, December 1972. 2. Hudson, D. E., "Strong Motion Seismology", Proc. Int. Conference on Microzonation, Seattle, 1972. 3. Hudson, D. E., "Strong-Motion Earthquake Measurements in Epicentral Regions", Proc. Sixth World Conference on Earth• quake Engineering, New Delhi, 1977. 4. Hudson, D. E., "Reliability of Records - Panel 1, Earthquakes", Proc. Sixth World Conference on Earthquake Engineer• ing, New Delhi, 1977. 5. Trifunac , M. D., "Surface Motion of a Semi-Cylindrical Alluvial Valley for Incident Plane SH Waves", Bull. Seis. Soc. Amer. , Vol. 61, No. 6, December 1971. 6. Trifunac, M. D., and Brady, A. G., "On the Correlation of Seismic Intensity Scales with the Peaks of Recorded 118
PARKFIELD I966 TEMBLOR-S25W M 5.6 EL CENTRO I940-SOUTH M 6.7
PARKFIELD 1966 NO. 2 - N65E
SAN FERNANDO I97I PACOIMA DAM -N76W
M 6.4 f.O- "CP -z.' 0.8- PARKFIELD I966 NO 5-N85E p 0.6-
g 0.4-
^ 0.2
< 0^ PARKFIELD 1966 NO. 8-N50E
OLYMPIA I949-N86E M7.I
HELENA I935 WEST M 6.0 K STONE CANYON I972 MELENDY RANCH-N 29 W
M 4.6
KOYNA I967-L0NG. GOLDEN GATE PARK - S80E STATE BUILDING - S09E M 6.5 SAN FRANCISCO I957
M 5.3
l^iW MANAGUA 1972- EAST I ( M 6.2 PORT HUENEME I957 -SOUTH
LYTLE CREEK I970-S65E M 4.7
M 5.4
0 5 10 15 20 25 30
TIME - SECONDS
ANCONA 1972 FIGURE 2: NEAR-FIELD ACCELEROGRAMS - ROCCA NS M 4.9 EARTHQUAKE GROUND ACCELERATIONS IN EPICENTRAL REGIONS (REF. 3) 119
PACOIMA DAM l.25g ALLUVIUM SITE ROCK SSTE
(LARGER HORIZONTAL COMPONENT OF PEAK GROUND ACCELERATION)
DISTANCE, km FIGURE 3: PEAK HORIZONTAL GROUND ACCELERATIONS - SAN FERNANDO, CALIFORNIA, 1971 EARTHQUAKE (REF.
FIGURE 5: SCHEMATIC DIAGRAM SHOWING EARTHQUAKE SOURCE REGION,
TRANSMISSION PATH, AND LOCAL SITE CONDITIONS (REF. 8)
FOURIER HMPLITUDE SPECTRUM OF ACCELERATION SMOOTHED SPECTRA OF GROUP 111 ACCELEROGRAMS.SOUTH COMFONENT ORTA HAS 8EEN INSTRUMENT CORRECTED
EPICENTER STRONG MOTION LOCATION MAP GROUND MEASUREMENTS PASADENA AND THE SAN FERNANDO EARTHQUAKE EPICENTER » SEISMOSCOPE STATION • SEISMOSCOPE - GUTENBERG O ACCELEROGRAPH ® ACCELEROGRAPH + SEISMOSCOPE
FREQUENCY - CPS FIGURE 4: DISTRIBUTION OF STRONG GROUND MOTION MEASUREMENTS IN PASADENA AREA - SAN FERNANDO, FOURIER AMPLITUDE SPECTRUM OF ACCELERATION SMOOTHED SPECTRR Of GROUP I ACCELERDGRRMS.SOUTH COMPONENT CALIFORNIA, 1971 EARTHQUAKE (REF. 1} INSTRUMENT CORRECTEO
2a P A ~i"U^"5'
/ \ X -- WAVELENGTH /90° \ OF INCIDENT / I SH WAVES ^0° / \ /I D V > NORMALIZE D >T SURFAC E DISPLACEME I K XiTVjK^eo" 5 "—^ go0 '
f SURFACE FREQUENCY - CP!
VALLEY ACCELEROGRAMS AT THE EL CENTRO, CALIFORNIA, ^. = RIGIDITY SITE (REF. 8) yy//Oy/y^ $ = SHEAR WAVE VELOCITY INCIDENT L SH WAVES ' x- ^ FIGURE 6: SURFACE MOTION OF A SEMI-CYLINDRICAL ALLUVIAL VALLEY FOR INCIDENT SHEAR WAVES (REF. 5) 120 VERTICAL VERTICAL o VERTICAL HORIZONTAL HORIZONTAL • HORIZONTAL I 1, m 12 2 21 sn 2m x xt EE XE II TTT TV V Vf W 2fll DC X XI XE i n HI E 2 si 3zn 221 ix x II m M M. INTENSITY MM. INTENSITY M. M. INTENSITY FIGURE 8: CORRELATION OF PEAK GROUND MOTIONS WITH MODIFIED MERCALLI INTENSITIES (REF. 6) H0RI2 VERT H0R1Z VERT HORIZ VERT 0[ I 2 ~o\ I \Z 0 12 0 I 2 1 HE 0 12 0 12 2 SITE o| SITE SITE 'I CLASSIFICATION CLASSIFICATION CLASSIFICATION 0 - SOFT 0-SOFT 0-SOFT 1 - INTERMEDIATE I - INTERMEDIATE 1 - INTERMEDIATE -2 - HARD 2 - HARD 2 - HARD ii M.M. INTENSITY M.M. INTENSITY M.M. INTENSITY FIGURE 9: CORRELATION OF PEAK GROUND MOTION WITH INTENSITIES AND SITE CONDITIONS (REF. 6)