Global Seismic Network data were used to image upper-mantle seismic discontinuities. Stacks of phases that precede the PP phase, thought to be underside reflections from the upper-mantle discontinuities at depths of 410 and 660 kilometers, show that the reflection from 410 kilometers is present, but the reflection from 660 kilometers is not observed. A continuous Lame's constant lambda and seismic parameter at the 660-kilometer discontinuity explain the missing underside P reflections and lead to a P-wave velocity jump of only 2 percent, whereas the S-wave velocity and density remain unchanged with respect to previous global models. The model deemphasizes the role of Lame's constant lambda with regard to the shear modulus and constrains the mineralogical composition across the discontinuity.

The St. Elias, Alaska earthquake of 28 February, 1979 (M s 7.2) is reanalyzed using broadband teleseismic body waves and long-period surface waves because of unresolved questions about its depth, focal mechanism, seismic moment, and location in a seismic gap. Teleseismic waveforms are simultaneously inverted to determine the source mechanism, seismic moment, rupture history and centroid depth. These data are well modeled with a point source propagating in the ESE direction with an average kinematic rupture velocity of 2.5 km/s. The best-fitting source mechanism indicates underthrusting on a NE-dipping plane. The mainshock depth of 24 km and the depth of aftershocks determined from inversions are consistent with locations on the gently dipping main thrust of the Pacific-North American plate boundary. These depths are substantially different from those of earlier body wave studies and regional seismic network aftershock depth determinations but are in accord with the Harvard Centroid-Moment Tensor and International Seismological Centre determinations. The seismic moment determined from body waves is 9.4x1019N-m (M w 7.3). The spatial and temporal distribution of moment release indicates that the St. Elias earthquake was a complex rupture consisting of two distinct subevents within 38 s of the initial onset, followed by low moment release during the next 34 s. Earlier studies indicated an unusual amount of surface wave energy at very long periods (> 200 s) that led some workers to suggest that St. Elias was a "slow" earthquake. Our broadband modeling does not require more than 34 s of additional moment release after the first two subevents. Moreover, we are able to match the phase and amplitude of 200-s Love and Rayleigh waves with a thrust fault point source of moment 1.3x102ø N-m (M w 7.4) located at the body wave centroid. The moment difference is not discernible with body waves for moment evenly distributed over 72 s. Thus, the St. Elias earthquake is not slow with respect to 200-s surface waves but is complex with regard to the broadband body waves. Upper plate structure apparently controlled the gross characteristics of rupture. The rupture direction parallels mapped upper plate faults. Rupture propagated unilaterally to the ESE, with little initial moment release, as a shallow, north-dipping thrust that later changed to more steeply NE dipping with a large right-lateral strike-slip component. The locations and source mechanisms of these subevents and locations of aftershocks define a shallow dipping surface at the eastern edge of the Pacific plate. Moreover, the component of strike-slip motion increases with time in the mainshock implying that the transition to strike-slip faulting occurs along the plate interface. The estimated nucleation point of the second subevent coincides with a large concentration of aftershocks interpreted as representing a barrier to continuous rupture associated with the northern-most boundary of the Yakutat terrane. Joint relocation of aftershocks suggests that the main plate bound...

The 500‐km‐long Alaska Peninsula and Shumagin Islands segments of the Aleutian arc have a moderate to high probability of rupturing in one or more large or great earthquakes in the next few decades. To understand the likely modes of rupture in the next sequence of large and great events and to delineate the current geometry of the plate interface, we determine focal mechanisms, depths, and source time functions from seismic records for the largest events since 1917: the great earthquake of November 10,1938, and seven events with surface wave magnitude (Ms) of 6.9 to 7.5. Teleseismic body waves and surface waves are used to estimate the seismic moment and gross rupture characteristics of the great earthquake of November 10,1938, along the Alaska Peninsula. Body wave inversion of five P and four SH waves gives a duration of about 110 s in which moment was released in two episodes, each of about 50 s duration, with the second being larger than the first. The first source was located in the general vicinity of the epicenter of the mainshock, and the second, which occurred about 60 s later, was centered about 180 km to the northeast. The body‐wave‐derived seismic moment is 3.7×1021 N m (Mw 8.3). We corroborate the body wave results by calculating surface waves by normal‐mode summation and comparing them with data at periods greater than 50 s. An adequate fit to observed seismograms is obtained for either a single point source or two point sources, with one located about 180 km northeast of the mainshock epicenter. Rupture in 1938 appears to have been confined to the Alaska Peninsula segment; uniform rupture into the Shumagin region is not supported by the data. The Ms 6.9 earthquake of May 13,1993, ruptured a small portion of the Shumagin gap. The earthquake of May 14,1948, (Ms 7.5) occurred on a shallow dipping thrust fault with a depth of about 31 km, not 60 km as originally suggested. Five Ms ∼ 7 events in the 1938 rupture zone have locations, depths, and mechanisms that define a shallow dipping (16°–19°) plate interface which shallows to 8° under Kodiak Island. Subduction of seamounts of the Gulf of Alaska seamount province may explain the location, moment, focal mechanism, and depth of the five Ms ∼ 7 events. The plate interface dips nearly uniformly between the Alaska Peninsula and Shumagin segments, indicating that segmentation, if any, is not controlled by the orientation of the plate interface. A slight warp of the plate interface may form the boundary between rupture zones of great earthquakes along the Kodiak and Alaska Peninsula segments of the arc and may explain why rupture zones of large to great earthquakes rarely cross this tectonic boundary.

Summary Source parameters of the 1996 Flores Sea and 1994 Fiji–Tonga deep earthquakes are derived from teleseismic body waves recorded by the global seismic network of broad‐band seismograph stations. Both events consisted of several subevents. Models to approximate the spatial and temporal extent of the source process include point sources, propagating point sources and a combination of these. For the Flores Sea event, rupture lasted about 23 s and terminated some 70 km east of the nucleation point as inferred from the duration of P‐wave pulses, in agreement with the findings reported by other investigators. Our preferred model suggests bilateral rupture propagation. It consists of four point sources that have variable double‐couple radiation patterns and source time histories, and explains well the large compensated linear vector dipole (CLVD) component inferred from the Harvard centroid moment tensor (CMT) solution. The main moment release in the Fiji–Tonga event lasted only about 15 s. Our best model consists of two point sources with a total moment release of 2.8 × 1020 N m. Rupture propagated subhorizontally from the nucleation point to the north. The termination of rupture was located about 40 km to the north of rupture initiation. The inferred velocity of moment release in the Flores Sea and Fiji–Tonga events was 3–5 km s‐1, a value which is higher than that inferred for the great Bolivian earthquake of June 1994. Other derived source parameters (static stress drop, radiated seismic energy and maximum seismic efficiency) are also significantly different from those inferred for the 1994 Bolivian event, suggesting that deep earthquake processes do not follow an easily detectable common mechanism.

Abstract. A comprehensive set of teleseismic waveforms from two South American deep-focus earthquakes of the predigital era, the 1970 Colombia (Mw-8.1) and 1963 Peru-Bolivia (Mw = 7.7) events, are inverted for source mechanism, seismic moment, rupture history, and cemroid depth. The P and SH wave inversion of the Colombia event confirms previous work, indicating that rupture occurred on a plane that dips steeply west. Rupture direction paralleled the trend of the Wadati-Benioff zone. We decompose the source into subevents, based on a source time function which shows two major moment release pulses separated by ~20 s. The first subevent is located near the initiation point at a depth of ~630 kin. The main moment release was located ~70 km to the southeast and ~20 km shallower. Rupture subsequently propagated farther southeast.