CUREe Proceedings for Meeting in Los Angeles (August 20-22, 1997)

Estimation of Uniformly Spaced, Near-Source, Broadband Ground Motions for the 1994 Northridge Earthquake From Forward and Inverse Modeling

DAVID J. WALD, U.S. GEOLOGICAL SURVEY

PASADENA, CALIFORNIA

KENNETH W. HUDNUT, U.S. GEOLOGICAL SURVEY

PASADENA, CALIFORNIA

THOMAS H. HEATON, CALTECH

PASADENA, CALIFORNIA

  • Download 144 Simulated time histories (1.8 Mb gzipped)

    • SUMMARY

    We have generated realistic broadband ground motions that represent the range and variability of the near-source ground motions from the Northridge earthquake. We augment the relatively sparse recorded data set with evenly spaced sites for ground motions computed using a hybrid construction of simulated and observed recordings. In essence, we are estimating ground motions at many sites that did not record the mainshock by inferring the motions based on our model of the Northridge rupture (Wald et al., 1996). The final product of this study consists of 144 three-component, broadband ground motion time histories that can be used as input into engineering design and hazards assessment. This allows for analysis of the damage potential of the ground motions over the entire near-source region, rather than relying on poorly spatially-sampled data alone. These simulated ground motions have been used by earthquake engineers in the U.S. and Japan for building response analyses and are available for similar or other uses.

    THE NEED FOR BROADBAND GROUND MOTION SIMULATIONS

    The recent awareness of the importance of coherent, long-period velocity pulses in near-source ground motions, in addition to well-studied strong-ground accelerations, have sparked concern about the response of engineered structures to such motions. This concern was, in large part, a result of widespread damage and severe ground-motion recordings from the 1994 Northridge earthquake, and has been rekindled by the death and destruction caused by the 1995 Kobe, Japan earthquake. The numerous near-source Kobe and Northridge ground-motion recordings reaffirm the potential severity of long-period velocity pulses which have been noted from records of previous earthquakes. These velocity pulses can result in very high drift demands in flexible structures (Hall et al., 1995) and indeed, a surprising number of welded beam-column connections failed during the Northridge earthquake--most in structures well away from the regions that experienced the largest ground velocities (Wald et al., 1996). Despite being discussed in the seismological community for some time, (e.g., Archuleta and Hartzell, 1981), such motions have traditionally been poorly represented as input into engineering design.

    Although there are relatively few near-source recordings of ground motions from large earthquakes, the records that are available show large velocities (see Heaton et al., 1995). One striking characteristic is the large, coherent pulse caused by the passing of the rupture front (source directivity). These features can be predicted for specific source geometries with finite fault rupture models. Recent finite-fault source studies of the 1989 Loma Prieta (Wald et al., 1991), the 1992 Landers (Wald and Heaton, 1994), and the 1994 Northridge (Wald et al., 1996) earthquakes have provided a documented calibration for simulations shown in this study.

    There is now an intense interest in simulating realistic broad-band ground motions from large earthquakes which contain these longer-period, directivity pulses and also realistically represent higher frequency accelerations. The simulated motions are required for input into engineering design and for the evaluation of existing structures to such earthquake loads. Previous studies have shown that ground motions can be accurately calculated deterministically for periods longer than 1-2 sec., and near-source ground motions have been successfully modeled for at least ten California earthquakes (see Heaton et al., 1995, for a list of earthquakes and references). At periods shorter than 1-2 sec, it is difficult to produce deterministic ground motions and typically empirical (e.g., Hartzell, 1978; Kanamori et al., 1988), semi-empirical (Irikura, 1983; Wald et al., 1988; Somerville et al., 1991), stochastic methods (e.g., Boore, 1983; Silva et al., 1990), or hybrid techniques (Heaton et al., 1995; Zheng et al., 1994) are used. In our approach to modeling broadband strong motions, we employ the finite fault methodology of Hartzell and Heaton (1983) for the long-period deterministic ground motions, and we chose to use actual recordings of accelerations to represent high frequency energy. Specifically, we use a matched pair of filters to remove short periods from our synthetics and to remove long periods from the actual recorded ground motions.

    The filtered records are then summed to form the final, broadband ground motions. In this way, those earthquake-specific ground-motion characteristics that can currently be modeled in a deterministic manner (long-period wave propagation, slip distribution, source geometry, and source radiation) are done so with existing methodologies, and the higher frequency effects are added in empirically.

    Click here for Picture

    Figure 1. The slip distribution derived from the inversion of strong-motion, teleseismic and geodetic data. Contour interval is 0.4 meters. The hypocenter is given with a star. View is above and perpendicular to the fault plane as viewed from the south-southwest.

    We refer to this approach as "infilling" or estimating broadband ground motions for the entire near-source region of a well-studied earthquake, in this case the 1994 Northridge event, by ''filling in'' where no data were recorded. From the source rupture model of Wald et al. (1996), and using the procedure match-filtering procedure outlined here, we predicted the ground motions on a grid of stations above the rupture surface to augment the sparse recordings made in this region. This is an effective way of estimating the overall damage pattern and potential of an event like the Northridge earthquake as well as focusing in on more site-specific ground motion and design issues.

    Click here for Picture

    Figure 2. Observed horizontal ground velocities (oriented to maximize peak-to-peak amplitude) shown with the time series origin at the station location. The basemap is southern California topography, with basins darkly shaded. The epicenter is given with a star and the fault projection of the fault plane to map view is shown as a dashed rectangle. Both time and amplitude scales are shown in the upper right.

    SLIP DISTRIBUTION INVERSION RESULTS

    We previously determined a rupture model for the 1995 Northridge, California earthquake from the joint inversion of near-source strong ground motion recordings, P and SH teleseismic bodywaves, Global Positioning System (GPS) displacement vectors, and permanent uplift measured along leveling lines. Effectively, this results in the pattern and timing of slip on the rupture plane that best matches these data given the particular set of assumptions about the faulting geometry, the rupture parameterization, and the earth structure.

    The fault is defined to strike 122[ring] and dip 40[ring] to the south-southwest consistent with the above data sets and the faulting geometry inferred from the pattern of aftershock locations. The average rake vector is determined to be 101[ring] (nearly pure thrusting) and the average slip is 1.3 meters (Fig. 1); the peak slip reaches about 3 meters. Our estimate of the seismic moment is 1.3 +/- 0.2 x 1026 dyne-cm (seismic potency, slip times area, of 0.4 km3). The area of significant slip is small relative to the overall aftershock dimensions and is approximately 15 km along strike, nearly 20 km in the dip direction, with no indication of slip shallower than about 5-6 km (Fig. 1).

    The up-dip, strong-motion velocity waveforms are dominated by large S-wave pulses attributed to source directivity and are comprised of at least 2-3 distinct arrivals (a few seconds apart) as seen in Figure 2. Stations at southern azimuths indicate two main S-wave arrivals separated slightly longer in time (about 4-5 sec). These observations are best modeled with a complex distribution of subevents: The initial S wave arrival comes from an asperity that begins at the hypocenter and extends up-dip and to the north where a second, larger subevent is centered (about 12 km away, Fig. 1). The secondary S arrivals at southern azimuths are best fit with additional energy radiation from another high slip region at a depth of 19 km, 8 km west of the hypocenter.

    Both the observations and the rupture model indicate that the strongest long-period (1 to 3 sec) ground motions occurred up-dip from the rupture surface where source directivity is greatest. We note that much of this same region is sparsely populated and has few, if any, larger steel- or concrete-frame structures. Consequently, the engineering problems discovered in these structures after the earthquake occurred at relatively modest levels of ground velocity relative to those motions experienced north and northeast of the epicenter (Wald et al., 1996).

    FORWARD BROADBAND GROUND MOTION ESTIMATES

    After determining the overall spatial and temporal variations of slip on the Northridge rupture plane, we chose a grid of target sites in the near-source area, with 144 stations covering an area of approximately 3600 square km (station spacing about 5 km). Next, long-period (> 1 sec) ground motions were computed for each grid station. The grid of stations used is shown in Figure 4 with small circles; open circles depict rock sites and closed symbols represent soil sites. Green's functions at rock and soil sites were computed with two different velocity structure models (taken from Wald et al., 1996) to better approximate the shallow impedance contrasts. Short-period (< 1 sec) Click here for Picture

    Figure 3. Observed peak ground velocities contoured in intervals of 20 cm/sec. The basemap is southern California topography, with basins darkly shaded. Diamond symbols denote strong motion stations.

    TABLE 1

    Peak Ground DiSPLACEMENT, VELOCITY and ACCELERATION FOR THE HORIZONTAL COMPONENT AT 100 SITES OF THE NORTHRIDGE EARTHQUAKE SIMULATION

    PEAK GROUND DISPLACEMENT (cm)


    		2
    		3
    		4
    		5
    		6
    		7
    		8
    		9
    		10
    		11
    J
    		7.7
    		7.7
    		10.2
    		8.9
    		11.2
    		14.9
    		12.3
    		13.9
    		7.5
    		5.2
    I
    		9.0
    		12.4
    		16.5
    		15.9
    		36.4
    		28.3
    		14.3
    		11.0
    		8.0
    		4.8
    H
    		20.4
    		32.9
    		52.3
    		29.8
    		60.8
    		37.2
    		12.6
    		9.4
    		6.7
    		5.6
    G
    		8.0
    		19.5
    		16.7
    		31.1
    		30.1
    		35.3
    		21.2
    		7.1
    		4.2
    		4.3
    F
    		7.6
    		17.1
    		20.5
    		17.2
    		10.2
    		21.5
    		47.9
    		20.6
    		5.5
    		3.9
    E
    		9.2
    		20.9
    		19.9
    		9.2
    		30.4
    		29.1
    		28.8
    		25.9
    		10.6
    		5.7
    D
    		7.6
    		10.3
    		10.3
    		8.5
    		25.5
    		20.4
    		18.3
    		14.8
    		3.6
    		3.2
    C
    		6.4
    		9.9
    		9.6
    		28.3
    		16.7
    		14.1
    		9.0
    		4.8
    		5.8
    		3.9











    PEAK GROUND VELOCITY (cm/sec)
    J
    		29.
    		35.
    		38.
    		52.
    		46.
    		50.
    		38.
    		39.
    		21.
    		13.
    I
    		32.
    		30.
    		42.
    		64.
    		103.
    		81.
    		37.
    		42.
    		14.
    		16.
    H
    		58.
    		83.
    		176
    		83.
    		175.
    		117.
    		54.
    		29.
    		45.
    		42.
    G
    		23.
    		35.
    		37.
    		37.
    		50.
    		58.
    		54.
    		34.
    		35.
    		43.
    F
    		24.
    		28.
    		35.
    		35.
    		44.
    		46.
    		118.
    		81
    		34.
    		37.
    E
    		19.
    		64.
    		65.
    		33.
    		123.
    		77.
    		95.
    		73.
    		27.
    		26.
    D
    		24.
    		33.
    		29.
    		22.
    		68.
    		82.
    		41.
    		43.
    		27.
    		17.
    C
    		26.
    		32.
    		32.
    		93.
    		68.
    		43.
    		33.
    		32.
    		35.
    		26.











    PEAK GROUND ACCELERATION (cm/sec2)
    J
    		187.
    		237.
    		223.
    		262.
    		289.
    		274.
    		488.
    		424.
    		130.
    		135.
    I
    		213.
    		210.
    		249.
    		312.
    		531.
    		727.
    		333.
    		444.
    		149.
    		138.
    H
    		326.
    		445.
    		754.
    		334.
    		854.
    		774.
    		450.
    		432.
    		460.
    		448.
    G
    		34.
    		144.
    		203.
    		189.
    		199.
    		318.
    		307.
    		432.
    		424.
    		446.
    F
    		258.
    		170.
    		193.
    		279.
    		186.
    		246.
    		743.
    		784.
    		431.
    		450.
    E
    		164.
    		523.
    		540.
    		238.
    		690.
    		634.
    		499.
    		600.
    		222.
    		347.
    D
    		185.
    		239.
    		221.
    		221.
    		481.
    		437.
    		310.
    		330
    		312.
    		173.
    C
    		197.
    		239.
    		238.
    		499.
    		443.
    		420.
    		334.
    		312.
    		357.
    		214.











    ground motions were then chosen from the observed Northridge recordings so as to best approximate the correct directivity, site conditions, and epicentral distance for each of the grid stations. Finally, a matched pair of filters was used to remove the short periods from the computed, deterministic recordings and to remove the long periods from the actual recordings, and these two filtered records were summed to produce the final ground motion.

    Figure 4 summarizes the overall pattern of peak velocities contoured over the computed grid. The peak ground displacements, velocities, and accelerations for 80 of the near-source grid stations nearest the source are tabulated in Table 1, with the horizontal components rotated to maximize the peak velocity. For the horizontal components, the largest acceleration is 854 cm/sec2 at site H06; the largest velocity is 176 cm/sec (69 in/sec) at H04 and the largest displacement is 61 cm (24 inches) at site H06. Examples of the simulated acceleration, velocity and displacement broadband time histories are given in Figure 5 along with spectral response. The largest ground velocities produced in the simulation are due to near-source effects, but they are also at soil sites where significant amplification occurs relative to rock sites. Rupture starts at the hypocenter and propagates updip and to the north. The regions where the extension of the buried thrust fault intersects the ground surface is where the most notable near-source directivity effects are found.

    Click here for Picture

    Figure 4. Predicted peak ground velocities contoured in intervals of 20 cm/sec. The basemap is southern California topography, with basins darkly shaded. Circles denote grid station locations; filled circles represent soil sites; unfilled represent rock sites.

    Note that the overall contour pattern of the predicted peak ground velocities closely resemble the observed pattern (Fig. 3) as it should; these observed ground velocities were used explicitly in a least-squares inversion for the slip history on the fault. However, the contours of the observed ground velocity are based on a irregularly spaced, coarsely-sampled observations, so the pattern is spatially aliased. In contrast, the predicted pattern is generated from a more finely sampled, uniform grid of stations and therefore it more adequately represents the ground velocity patterns in areas without observations. We have not predicted ground motions out into the Los Angeles basin since this requires further validation with respect to 3D wave propagation.

    The ground motion time series generated in these simulations can be obtained from the authors, or they can be directly downloaded in ASCII formatted files from the World Wide Web at URL http://www-socal.wr.usgs.gov/wald/CUREe.html.

    Click here for Picture

    Figure 5. Simulated acceleration, velocity, displacement and response spectra (5% damping) for grid stations F08 and E06.

    REFERENCES

    Archuleta, R. and S. Hartzell (1981). Effects of fault finiteness on near source ground motion, Bull. Seism. Soc. Am. , vol. 71, 939-957.

    Boore, D. M. (1983). Stochastic simulation of high-frequency ground motions based on seismological models of the radiated spectra, Bull. Seism. Soc. Am. , vol. 73 , 1865-1894.

    Hall, J. F., T. H. Heaton, M. W. Halling, and D. J. Wald (1995). Near-source ground motions and its effects on flexible buildings, Earthquake Spectra , vol. 11, 569-606.

    Hartzell, S. H. and T. H. Heaton (1983). Inversion of strong ground motion and teleseismic waveform data for the fault rupture history of the 1979 Imperial Valley, California earthquake, Bull. Seism. Soc. Am., vol. 73, 1553-1583.

    Hartzell, S. H. (1978). Earthquake aftershocks as Green's functions, Geophys. Res. Lett., vol. 5 , 1-4.

    Heaton, T. H., J. F. Hall, D. J. Wald, and M. W. Halling (1995). The response of high-rise and base-isolated buildings to a hypothetical Mw = 7.0 blind thrust earthquake, Science, vol. 267, 206-211.

    Irikura, K. (1983). Semi-empirical estimation of strong ground motions during large earthquakes, Bull. Disast. Res. Inst. , Kyoto Univ., vol. 33, 63-104.

    Kanamori, H., Jennings, P. C., D. Helmberger, R. Clayton, J. K. Singh, L. Astiz, and E. Mena (1988). Estimation of ground motions in Mexico City, Proceedings of Ninth World Conference on Earthquake Engineering, vol. VIII, 43-49.

    Silva, W., R. Darragh, C. Stark, I. Wong, J.C. Stepp, J. F. Schneider, and S-J Chiou (1990). A methodology to estimate design response spectra in the near-source region of large earthquakes using the band-limited-white-noise ground motion model, 4th U. S. Nat'l Conf. on Earthquake Engr., vol. 1, 487-494.

    Somerville, P. G., M. K. Sen, and B. Cohee (1991). Simulation of strong ground motions recorded during the 1985 Michoacan, Mexico and Valparaiso, Chile earthquakes, Bull. Seism. Soc. Am., vol. 81, 1-27.

    Wald, D. J., L. J. Burdick, and P. G. Somerville, (1988). Simulation of acceleration time histories close to large earthquakes: Proceedings of the Earthquake Engineering and Soil Dynamics II Conference, American Society of Civil Engineers, Geotechnical Special Publication No. 20, 430-444.

    Wald, D. J., T. H. Heaton and D. V. Helmberger (1991). Rupture model of the 1989 Loma Prieta earthquake from the inversion of strong motion and broadband teleseismic data, Bull. Seism. Soc. Am., vol. 81, 1540-1572.

    Wald, D. J., and T. H. Heaton (1994). Spatial and Temporal distribution of slip for the 1992 Landers, California, earthquake, Bull. Seism. Soc. Am., vol. 84, 668-681.

    Wald, D. J., T. H. Heaton, and K. W. Hudnut (1996). The slip history of the 1994 Northridge, California, earthquake determined from strong-motion, teleseismic, GPS, and Leveling data, Bull. Seism. Soc. Am. , vol. 86, S49-S70.

    Zheng, Y., J. G. Anderson, and G. Yu (1994). A composite source model for computing realistic synthetic strong ground motions, Geophys. Res. Lett., vol. 21 , 725-728.