In Chapter 1, we analyzed short-period body waves recorded at teleseismic distances from great earthquakes. They provide information about source rupture processes and strong motions. First, we examined records of 19 earthquakes of moment magnitude MW of 6.5 to 9.5. Four parameters were measured from the short-period P-wave train: the maximum amplitude, the period at maximum amplitude, the time between the first arrival and when the maximum amplitude is attained, and coda length. An extension, m̂b, of the teleseismic body-wave magnitude, mb, was defined using the maximum amplitude of the entire short-period P wave rather than the amplitude achieved in the first few P-wave cycles. m̂b increases as MW increases. The time from the first arrival until the maximum amplitude is achieved and the coda length are roughly proportional to MW, but were further interpreted by a simple asperity model of the rupture process. These data support that 1 to 2 sec waves are on average generated preferentially in the same regions of the fault plane as 10 to 50 sec waves.
An application to strong motion modeling is presented in which a 1971 San Fernando teleseismic short-period record is summed up to simulate teleseismic records produced by five great earthquakes. The summation procedure matches the moment of the event to be simulated, and includes rupture propagation, fault plane roughness, and randomness. The m̂b data provide an important constraint on the summation procedure. Thus constrained, this summation procedure can be more confidently used with near-field strong motion records as Green's functions to predict strong motions from great earthquakes.
In Chapter 2, we analyzed the spectra of short- and intermediate-period teleseismic GDSN (digital) records for seven large earthquakes and hand-digitized short-period analog records of the 1971 San Fernando earthquake. To obtain the source or moment-rate spectrum at periods between 1 and 30 sec, we Fourier-transformed the P waves, corrected for instrument response, attenuation, geometrical spreading, and radiation pattern (including the depth phases), and then averaged the records for each event. Significant differences exist between the spectra of different events, presumably due to variations in tectonic setting or seismic coupling. Using the digital data, we also investigated the relationship between time-domain amplitude and spectral amplitude for short-period P waves. From our empirical relation between spectral amplitude and time-domain amplitude, we estimated the spectral amplitudes implied by the m̂b data. We compared our results to the ω-2 and Gusev source spectral models. Neither model can completely represent the data. Nevertheless, we consider the ω-2 model a useful reference model for comparing different events. The average source spectrum of six large events with MW's of 7.4 to 7.8 does not have the spectral structure suggested by Gusev.
In Chapter 3, source characteristics of the Sept. 19, 1985 Michoacan, Mexico earthquake and its aftershock on Sept. 21 are inferred from broadband and short-period teleseismic GDSN records. We obtained the teleseismic source spectrum from 1 to 30 sec. The Michoacan source spectrum is enriched at 30 sec and depleted at 1 to 10 sec relative to an average source spectrum of large interplate subduction events. Source spectra for the Sept. 21 aftershock, 1981 Playa Azul, 1979 Petatlan, and 1978 Oaxaca events follow a trend similar to that of the 1985 Michoacan event. This spectral trend may characterize the Mexican subduction zone.
A station-by-station least-squares inversion of the Michoacan earthquake records for the source time function yields three source pulses, which we interpreted as events on the fault plane. The first two are similar in moment, and the third contains only 20% of the moment of the first. Directivity is evident in the timing. At each station, we measured the time differences between the pulses, and performed a least-squares nonlinear estimation of the strike, distance, and time separation between the events to locate them relative to one another. The second event occurred 26 sec after the first, and 82 km southeast of it, indicating southeastward rupture along the trench. The two large events are also seen in the near-field strong motions.
The mainshock records, spectrum, and time functions contain less high frequency radiation than those of the 1985 Valparaiso, Chile earthquake. Apparently, the Michoacan earthquake ruptured two relatively smooth, strong patches which generated large 30 sec waves, but small 1 to 10 sec waves. Such behavior contrasts with the Valparaiso event which had a more complex rupture process and generated more 1 to 5 sec energy. This difference is consistent with the higher near-field accelerations recorded for the Valparaiso event.
In Chapter 4, time functions and rupture processes of 4 recent large subduction zone earthquakes were determined from broadband teleseismic GDSN records using the iterative inversion technique of Kikuchi and Fukao. The method inverts the records simultaneously by determining the location, time, and seismic moment of a single point source that best explains the records, then subtracting the synthetics for that point source from the records, and repeating the procedure for the residual records. Using this technique with the high-quality GDSN intermediate-band records provides more details of the rupture process than have been obtained previously.
Using the inversion, we produced maps of the spatial and temporal distribution of seismic moment release at periods of 3 to 30 sec for the 1985 Michoacan, Mexico, 1983 Akita-Oki, Japan, 1985 Valparaiso, Chile, and 1986 Andreanof Islands, Alaska earthquakes. These are four of the largest earthquakes that occurred from 1983 to 1986. Comparing the source spectra of these events yields complementary information at periods of 1 to 30 sec. These four earthquakes have distinct rupture styles.
In Chapter 5, we compared strong motion spectra of the 1985 Michoacan, Mexico and the 1983 Akita-Oki, Japan earthquakes with their teleseismic spectra. The spectral levels of the Michoacan strong motions, which were recorded by a high-quality digital array, agree to within a factor of 2 with those predicted by the Michoacan teleseismic records. The Michoacan teleseismic spectrum is lower than that of Akita-Oki. This relationship also holds for their strong motion spectra. This consistency means that teleseismic records, which are relatively more abundant, can be used to predict properties of strong motions from large earthquakes.
