Search

Intermediate and deep focus earthquakes are generally thought to produce few aftershocks. The two year operation of a network of broadband seismographs in the Tonga-Fiji region provides an opportunity to monitor the aftershock activity of moderate-sized earthquakes. In addition to the March 9, 1994 (Mw 7.6) event that shows 144 aftershocks, we found that all nine intermediate and deep earthquakes with Mw > 6.0 showed some aftershock activity, suggesting that aftershock sequences are ubiquitous. On average, deep Tonga earthquakes show an order of magnitude fewer aftershocks above a given magnitude than shallow California earthquakes of similar moment. The largest aftershocks of each sequence have an mb that is 2.0 units smaller on average than the Mw of the mainshock, whereas the California earthquakes show a mean difference of 1.1. Since b-values of the deeper sequences appear to be similar to shallow sequences, the difference in aftershock productivity results from the larger size difference between the mainshock and the larger aftershocks. The most prolific sequences show an activity level not far below shallow earthquakes of the same moment. This is in contrast with observations of little aftershock activity for deep earthquakes in South America, Japan, and Kuriles, and suggests that the Tonga subduction zone is unusually prolific in aftershock production, perhaps due to cold slab temperatures.

We analyzed the seismic waves of the 1994 Sanriku-Haruka-Oki earthquake (Mw = 7.7), which occurred in the aftershock area of the 1968 Tokachi-Oki earthquake (Mw = 8.2). Applying a multiple-time window inversion scheme to near-source strong-motion data, we obtained a detailed spatiotemporal rupture process and compared it with that of the Tokachi-Oki earthquake. The fault geometry is constructed based on the aftershock distribution. The obtained rupture model is consistent with the CMT solution even for a non-double-couple component. The total seismic moment is 4.0 × 1020 N-m. Large slips are concentrated in three asperities: the first asperity centers about 40 km south and 50 km west from the hypocenter with a maximum slip of 4.4 m, the second one centers about 60 km west from the hypocenter with a maximum slip of 2.2 m, and the third one lies about 110 km west from the hypocenter with a maximum slip of 2.6 m. The obtained moment rate and the duration on the first and second asperities are lower and much longer than those on the third asperity, respectively. The first asperity does not overlap with an area of large slip during the Tokachi-Oki earthquake, but the second or third seem to overlap with or be adjacent to the asperity of the Tokachi-Oki earthquake. Our inversion result also shows an abrupt change of the rupture velocity (from 1.8 to 3.0 km/sec) at the central part of the fault plane. A difference of the seismic coupling between the oceanic and the continental lithospheres at the trenchward side and at the landward side of the 143° E meridian seems to affect the rupture process of this earthquake.

We use a numerical model for olivine-spinel transformation to study deep earthquake nucleation and to delineate the seismogenic region within a subducting slab. The model includes laboratory-derived flow laws, latent heat release, and phase transformation kinetics. We calculate deformation, transformation state, grain growth, and rheology for several paths within a subducting slab. Strain rate perturbations are imposed to define the necessary conditions for instability. Strain rate perturbations decay for ξ < a critical value ξc, and thus the coldest, interior portion of the metastable wedge deforms stably. For ξ≥ξc, strain rate perturbations grow, shear strength decreases with strain, and the system is potentially unstable. The instability condition is mapped to delineate the seismogenic zone within a subducting slab. The model seismogenic zone is bounded by ξc, and, at larger percent transformations, by coarsening of spinel grains and saturation of the transformation weakening effect. The model predicts a narrow seismogenic region along the outer edges of the metastable wedge and thus a “double seismic” zone, consistent with some seismic observations. Several simplifying assumptions are required due to lack of thermo-kinetic data and incomplete knowledge of constituent mineralogy and rheology. However, the model provides a quantitative definition of the instability condition and a framework for testing the hypothesis of transformation-induced instability.

We investigate the rupture process of the March 9, 1994, M w 7.6 deep Tonga earthquake and its relationship to the background seismicity of the subducted Tonga slab. Variations in observed P and S wave pulse duration indicate that the rupture propagated to the NNE and extended well beyond the background seismicity. We inverted 47 P and SH waveforms, including regional broadband waveforms from the Southwest Pacific Seismic Experiment, using a method that solves for the focal mechanism change during the rupture and the distribution of moment release along the fault plane. The results indicate that significant moment release occurred in previously aseismic regions outside the active seismic zone and that the rupture terminated 10-20 km beyond the bounds of the previous seismic activity. A significant change in focal mechanism occurred when the rupture propagated into the previously aseismic region. Rupture along the near-vertical NNE striking nodal plane provides a somewhat better fit to the body waveforms than rupture along the near-horizontal nodal plane. This result, combined with the planar alignment of aftershocks and the general NNE directivity of the waveforms, provides strong evidence that the rupture occurred on the near-vertical plane. Thermal modeling of the Tonga slab indicates that the rupture terminated in material about 200°C warmer than the temperature that normally limits the occurrence of smaller earthquakes. Additionally, aftershocks seem to be suppressed in the outer regions of the rupture, which contain about half of the moment release but only 1 of the 15 well-located aftershocks. We suggest that slabs may be composed of an inner cold core, where seismic rupture initiates and small earthquakes occur, and a thermal halo of warmer material, which can sustain rupture and only a few aftershocks. The mechanism by which rupture propagates through the warmer material need not be similar to the process governing rupture nucleation in the cold slab core; nucleation may occur through a process limited to the cold core such as transformational faulting, and propagation through the warmer material may occur through ductile faulting or plastic instabilities. Isolated deep earthquakes in other subduction zones, such as the 1994 Bolivia event, may occur almost completely within the warmer zone, accounting for the lack of background seismicity and the dearth of aftershocks.

The 1932 deep Fiji earthquake is studied from the point of view of comparing it to the recent large events in the transition zone. Spectral analysis and body-wave modeling suggest a moment of 3.4 × 10 27 dyn cm (within a multiplicative or divisive factor of 1.5), which makes it comparable to, but not significantly larger than, the great 1994 earthquake, 770 km to the north. Its seismic moment remains well below that of the largest deep shocks (1994 in Bolivia or 1970 in Colombia). Relocation efforts show that the earthquake took place on the front edge of the Wadati-Benioff Zone, in an area where the latter features a complex geometry involving significant warping. The mechanism of the earthquake involves down-dip compression and can be readily explained in terms of the prevailing large-scale stresses in the slabs.

From an analysis of broadband teleseismic P-waveforms we resolve the complex rupture characteristics of the 17 June 1996, Flores earthquake (Mw 7.9, depth 590 km). This earthquake was unusual because of its large rupture length and the variation in rupture velocity and stress drop. The 20 s rupture propagated westward at 2–3 km/s to a distance of 20–30 km, and eastward at 4 km/s over 75 km. Average static stress drop was about 15 MPa, but the stress drop during the episode of main moment release was a factor of four higher. The oblique normal faulting mechanism is consistent with the stress expected as the slab descends along a highly curved trench. Significant rupture complexity is also reported for the deep 1994 Mw 7.7 Fiji and Mw 8.2 Bolivia earthquakes and is evident in the source-time functions of many smaller earthquakes at depths >300 km. Thus source complexity of deep earthquakes appears no less common than for shallow earthquakes, and requires the existence of significant heterogeneity in the physical properties that control deep earthquake rupture.

We investigate the role of grain-size reduction during the olivine-spinal transformation on rheological properties of subducting slabs on the basis of a scaling model for micro structural development during nucleation and growth. In this model, the size of spinal grains nucleating at olivine grain-boundaries is controlled by the relative rates of nucleation and growth, taking into account the impingement through the collision of grains due to growth. When the volume fraction of spinal reaches a certain threshold value (critical volume fraction ~ 1 to 10 %, depending on the P,T conditions in the slab), the new phase will form a continuous film and will significantly reduce the strength of the two-phase aggregate if spinal grain-size is small. The size of spinal grains delta 0 at this stage is calculated and is shown to be highly sensitive to temperature. At relatively high temperatures (T > 1000 K), delta_0 shows an Arrhenius-type dependence on temperature, that is delta_0 ~ exp(-E* /RT) with E* ~ 400 kJ/mol, whereas a more complicated temperature dependence is found at low temperatures (T < 900 K), where a grain-size reduction of up to 4 orders in magnitude is possible. Strength profiles of slabs due to combined effects of temperatures and of grain-size reduction are calculated. It is shown that: (1) the strength of slabs will have an unusual temperature dependence through the temperature dependence of grain-size; and (2) a subducting slab has a complicated rheological structure containing a weak region below the tip of a metastable olivine-bearing wedge in a cold slab. Possible implications of these anomalous rheological structures of slabs on the dynamics of subduction are discussed including the mechanisms of deep earthquakes.

A constitutive model is developed to describe the creep behavior of olivine-spinel aggregates, and conditions for transformation shear instability and deep earthquake nucleation are predicted. Using the Eshelby-Mori-Tanaka theory [Eshelby, 1957; Mori and Tanaka, 1973], a piecewise incremental method is employed to calculate the stress-strain relation for the olivine-spinel aggregate. The creep strain and strain rate depend on differential stress, the aspect ratio of spinel inclusions, latent heat of the phase transformation, the relaxation of the volumetric phase transformation strain e ph kk , and the creep law parameters of spinel. A constant strain rate calculation suggests that above a critical percentage transformation ξ c an earthquake may nucleate owing to the system's potential for a shear stress drop, provided e kk ph of -8% is not relaxed and spinel grain size d xp remains small. A normalized critical stiffness, calculated from the potential shear stress drop Δτ and the critical shear strain γ c needed to change from one steady state creep condition to another, suggests that a plate-like, partially transformed region is required for deep earthquake nucleation. Such a region may be formed by localized deformation or preexisting heterogeneities in a subducting rock.

SUMMARY The ability of seismological criteria to identify earthquakes from underground explosions depends partly on the orientation of the earthquake source. Well-determined double-couple moment tensor solutions for a large number of earthquakes have been published in the Harvard centroid moment tensor (CMT) and United States Geological Survey (USGS) catalogues. Statistical analyses of these catalogues indicate that the distribution of the orientation of earthquake mechanisms is not random. The distribution of the T axes shows significant clustering around the downward vertical, indicating that a larger number of earthquake mechanisms radiate compressional P-wave energy to teleseismic distances from near the maximum of the radiation pattern than is predicted if earthquake sources are randomly oriented double couples. The clustered T axes correspond to compressional dip-slip mechanisms, and it is this type of mechanism which is believed to cause both the mb:kfs (the ratio of body-wave to surface-wave magnitude) and first-motion criteria to misidentify an earthquake as an explosion.

Short-period array seismograms of deep events that occurred in the Indonesia, Japan and Izu-Bonin arcs are stacked and beam-formed to identify the near-source S-P converted waves that result from the mantle transition discontinuities. Most of the resulting images reveal the existence of a mid-mantle seismic discontinuity (“920 km discontinuity”) in these regions. Of the 15 events analyzed, three that occurred at the western end of the Indonesia arc show clear S-P arrivals observable even in individual seismograms. The mid-mantle discontinuity is characterized by large depth variation (900 ∼ 1080 km) and velocity contrast variation in different subduction zones. Especially, the depth variation of the mid-mantle discontinuity beneath the Indonesia arc, where the discontinuity deepens from 940 km at the eastern end to 1080 km at the western end, appears to be well correlated with the location of the high-velocity anomalies in recent tomographic models. However, the mid-mantle discontinuity cannot be simply coincided with the bottom of the high-velocity anomalies, because a velocity increase at the discontinuity is observed from the waveform analysis.

We analyze P-wave codas from 12 deep Izu-Bonin earthquakes recorded by hundreds of Western United States seismograph stations for evidence of small-amplitude phases caused by near-source mantle discontinuities. For nearly every event, the dominant phase in the coda is the result of an S-to-P conversion from a nearly horizontal discontinuity ranging in depth from 650 to 690 km. This is interpreted as a thermally depressed spinel to perovskite and magnesiowustite phase transition. If the 660-km seismic discontinuity is also associated with a change in chemistry, it would be dynamically depressed by a subducting slab. We consistently see that there is no nearly horizontal discontinuity between 700 and 1000 km with shear wave velocity contrast exceeding 1%; this observation places constraints on the properties of a postulated chemical discontinuity separating the upper and lower mantles.

Using Wigner-distribution estimation, the high-frequency fall-off of source spectra was estimated. The technique was applied to the analysis of the source spectra of the Vladivostok M = 6.4 deep-focus earthquake of 22 July 1994 and the Northern Japan Sea, M = 5.7 deep-focus earthquake of 22 July 1994. It is shown that the high-frequency fall-off of the source spectra of these two earthquakes may be characterized by two parts. Between the first corner frequency f1 and the second corner frequency f2, the high-frequency fall-off of the source spectra can be represented by f−5; above f2, the high-frequency fall-off can be represented by f−2.

The distribution of earthquake seismic moment M is a Pareto (power-law) type with the exponent for shallow earthquakes. To ensure a finite total deformation energy, the distribution has to be modified for the largest earthquakes by introducing either a cutoff at or an exponential taper. Analysis of global seismicity suggests that both parameters do not change in a statistically significant way over continental areas and tectonic plate boundaries. Thus, very large earthquakes are possible in practically any urbanized areas. We present evidence that the statistical distribution of losses due to large earthquakes has a power-law (Pareto) tail with an exponent value less than 1.0: i.e., the distribution is heavy-tailed. If this statement is true, the earthquake average loss is controlled by the largest earthquakes. Neither the distribution of earthquake size at the maximum, nor the distribution of maximum losses is well known. This implies that insurance premiums can be determined only with significant uncert...

[1] Intermediate-depth earthquakes occur at depths where temperatures and pressures exceed those at which brittle failure is expected. There are two leading candidates for the physical mechanism behind these earthquakes: dehydration embrittlement and self-localizing thermal shear runaway. A complete energy budget for a range of earthquake sizes can help constrain whether either of these mechanisms might play a role in intermediate-depth earthquake rupture. The combination of high stress drop and low radiation efficiency that we observe for Mw 4–5 earthquakes in the Bucaramanga Nest implies a temperature increase of 600–1000°C for a centimeter-scale layer during earthquake failure. This suggests that substantial shear heating, and possibly partial melting, occurs during intermediate-depth earthquake failure. Our observations support thermal shear runaway as the mechanism for intermediate-depth earthquakes, which would help explain differences in their behavior compared to shallow earthquakes.

The sequence of phase transitions undergone by minerals with increasing depth in Earth's mantle is perturbed within subducting lithospheric slabs by their thermal structure. Such perturbation of equilibrium phase relations gives rise to relative buoyancy contrasts between slab and mantle that contribute to the state of stress within the slab. While other factors contribute to overall slab stresses, thermal and phase transition effects largely control the structure of the stress field within the slab. The resulting maximum in down-dip compressive stress within the slab corresponds to the observed peak in depth distribution of deep seismicity. Furthermore, metastable persistence of lower pressure phases within the cold slab should give rise to localized shear stresses whose distribution corresponds to observed features of subduction zone seismicity. These observations are independent of the variety of failure mechanisms proposed for deep seismogenesis.

A prominent and spatially localized phase arriving approximately 4 s after the initial compressional arrival, repeatedly observed on seismograms recorded in the central Rio Grande Rift from intermediate- and deep-focus earthquakes at distances near 90°, is identified as a probable compressional to shear wave conversion from a lower crustal root of the Socorro midcrustal magma body (SMB). This phase is inconsistent with previously determined crustal models containing a simple sill-like midcrustal magma body with a negligible conduit system. Finite element synthetic seismogram modeling suggests that a partially molten root with a western boundary dipping steeply to the east extends several kilometers from the 19-km-deep upper surface of the magma body down into the lower crust of the southwestern portion of the Albuquerque-Belen basin. The location and geometry of this feature suggest that the intracrustal intrusion of mantle magmas responsible for the ongoing inflation of the SMB and associated seismicity is occurring via an off-axis conduit system located beneath the active margin of the Albuquerque-Belen basin.

SUMMARY We determine the isotropic component of large deep earthquakes by inversion for their full six-component moment tensors using the normal-mode data. We show that it is possible to reduce the correlation between the isotropic component and other components and to determine the isotropic component independently by analyses of normal-mode data at periods longer than 500 s. We find no significant isotropic component for the earthquakes we studied; the magnitudes of the estimated isotropic components are comparable to the uncertainty due to the mislocation of the centroid. The magnitude of the isotropic component is at most 5 per cent of the deviatoric seismic moment if it exists. This suggests that, if a rapid phase change of mantle minerals occurs during the rupture of deep earthquakes, the mantle material that transformed would be confined within an extremely thin layer. We also conduct moment tensor inversion using body waves and surface waves in the period band between 20 and 500 s. We find that there is a relatively consistent non-double-couple component of the deviatoric moment tensor for the Japan Sea earthquake (1994 July 21). This is likely to be caused by source complexity

DEEP earthquakes generally show fewer aftershocks and a larger variability in magnitude–frequency relations than shallow earthquakes1–5. These characteristics and the factors that control them may place constraints on the mechanism of deep earthquakes, which is still uncertain6–8. Here we show that systematic variations in aftershock productivity and magnitude–frequency relations are related to the temperature of the downgoing slab, as inferred from the vertical velocity and age of the subducting lithosphere. Deep earthquakes with substantial aftershock sequences are found only in colder slabs, and earthquakes in warmer slabs consistently show few aftershocks. Magnitude–frequency relations for deep earthquakes also show systematic variations as a function of slab thermal structure, with warmer slabs showing fewer small earthquakes (lower 'b-values'). The statistical properties of deep earthquakes thus appear to be temperature sensitive to an extent not observed for shallow earthquakes, and not adequately explained by simple geometrical limitations on fault width.

We consider the possibility that the velocity structure of D″ is anisotropic. The data we examined consist of seismograms from 9 deep Japanese earthquakes recorded at WWSSN receiver stations in North America. The source-receiver combinations span distances of 70°–106° with associated S waves passing through D″ beneath Alaska. Differential travel times of the S, Scd, ScS and SKS phases are used to constrain the velocity structure in D″. Shear waves refracted by D″ are observed beyond 72.2° and provide a sensitive measurement of the velocity structure in D″. Beyond 93°, the vertically polarized (SV) and horizontally polarized (SH) shear waves often appear distinctly split, although, at distances less than 89° the components are more nearly synchronous. Near 94°, SH occurs as a double arrival. SV in this range, however, remains a single arrival roughly synchronous with the second SH arrival. We have been unable to reproduce these effects in isotropic model synthetics. Synthetics for transversely isotropic models have been computed that do match these waveforms. The anisotropy was constrained to be only within D″, with a vertical symmetry axis. We conclude that these observations may be explained by an anisotropic D″ layer. The D″ discontinuity may be due to a transition to anisotropic mantle a few hundred kilometers above the core-mantle boundary.

SUMMARY The observability of the isotropic component of a general moment tensor is investigated by carefully examining its resolvability from the other components. It is shown that the portion of seismograms between the first P-wave arrival and just before the arrival of the first surface wavetrain, which contains many different body-wave phases, can be used to constrain the isotropic component of deep earthquakes. CMT inversion using this portion of the seismogram for major deep earthquakes has revealed that there is no significant isotropic component for deep earthquakes. This conclusion contradicts some earlier studies, for example Dziewonski & Gilbert (1974), who found significant isotropic components of deep earthquakes by analysing normal-mode data. It is shown that the isotropic component of deep earthquakes cannot be resolved independently from the vertical CLVD component of the moment tensor by analysing the type of normal-mode data above 2mHz that previous researchers used. At the same time, however, the possibility of resolving the isotropic component using normal-mode data below 2 mHz is suggested.

An important constraint on the size and destructive potential of earthquakes is the depth extent of rupture. Laboratory studies of the transition from unstable to stable sliding, along with observed relationships between surface heat flow and the thickness of the seismogenic crust, provide strong evidence for the significance of temperature in determining the maximum nucleation depth of large earthquakes. The June 28, 1992, Mw 7.3 Landers earthquake ruptured fault segments within 20 km of 11 pre-existing heat flow measurements, and shallowing of the base of aftershock seismicity along strike correlates with an increase in heat flow. Crustal geotherms estimated from these measurements place the base of seismicity along the 250°C isotherm. This temperature is consistent with predictions from laboratory studies of the frictional stability of Westerly granite, but estimated temperatures for the seismic-aseismic transition along other faults within the San Andreas fault system are in the range of 350 to 400°C. Variations in country rock and fault gouge composition, together with higher slip rates, may account for this difference, although part of the Landers seismogenic crust might remain unruptured.

Fault geometry, depth, and slip distribution of the Sanriku-oki earthquake of December 28, 1994 (Ms 7.5) are estimated from seismic waveforms, geodetic measurements, and tsunami waveforms, and compared with those of the 1968 Tokachi-oki earthquake (Mw 8.2), the most recent large earthquake in the epicentral region. Seismic wave inversions indicate a shallowly dipping thrust type mechanism and the focal depth of 22–28 km, representing an underthrust event at the subduction interface. The source time function has an initial stage of 20 s followed by a main pulse of 30 s duration, a shape similar to that of the 1968 Tokachi-oki earthquake. Joint inversion of geodetic and tsunami data shows that the region of the largest slip, 1.7 m, corresponds to a region of relatively small slip for the 1968 Tokachi-oki earthquake. Furthermore, the joint inversion shows that the region of the largest asperity of the 1968 earthquake had essentially zero slip during the 1994 Sanriku-oki earthquake. These results indicate that the dominant asperities ruptured by the 1994 Sanriku and 1968 Tokachi-oki earthquakes are different. The result of the joint inversion also shows that the aseismic zone near the Japan trench was not ruptured by the 1994 earthquake.

SUMMARY In this paper we present revised locations and original focal mechanisms computed for intermediate and deep earthquakes that occurred within the Southern Tyrrhenian subduction zone between 1988 and 1994, in order to improve our knowledge of the state of stress for this compressional margin. In particular, we define the stress distribution within a large portion of the descending slab, between 40 and about 450 km depth. The seismicity distribution reveals a continuous 40-50 km thick slab that abruptly increases its dip from subhorizontal in the Ionian Sea to a constant 70" dip in the Tyrrhenian. We computed focal mechanisms for events with magnitudes ranging from 2.7 and 5.7, obtaining the distribution of P- and T-axes for many events for which centroid moment tensor (CMT) solutions are not available, thus enabling the sampling of a larger depth range compared to previous studies. We define three portions of the slab characterized by different distributions of P- and T-axes. A general down-dip compression is found between 165 and 370 km depth, whereas in the upper part of the slab (40-165 km depth) the fault-plane solutions are strongly heterogeneous. Below 370 km the P-axes of the few deep events located further to the north have a shallower dip and are not aligned with the 70" dipping slab, possibly suggesting that they belong to a separated piece of subducted lithosphere. There is a good correspondence between the depth range in which the P-axes plunge closer to the slab dip (-70") and the interval characterized by the highest seismic energy release (190-370 km).

We explore lowermost mantle structure beneath the Pacific with long‐period recordings of the seismic core phases SKS, SP_dKS, and SKKS from 25 deep earthquakes. SP_dKS and SKKS are anomalously delayed relative to SKS for lower mantle paths beneath the southwest Pacific. Late SP_dKS arrivals are explained by a laterally varying mantle‐side boundary layer at the CMB, having P‐velocity reductions of up to 10% and thickness up to 40 km. This layer is detected beneath a tomographically resolved large‐scale low velocity feature in the lower mantle beneath the central‐Pacific. SKS, SP_dKS, and SKKS data for the generally faster‐than‐average circum‐Pacific lower mantle are well‐fit by models lacking any such low‐velocity boundary layer. The slow boundary layer beneath the central Pacific may be a localized zone of partial melt, or perhaps a chemically distinct layer, with its location linked to overlying upwelling motions.

Shear-wave splitting observed in three-component broadband recordings of local earthquakes near southern Vancouver Island, British Columbia, provide the first constraints on shear-wave anisotropy in the continental crust above the subducting Juan de Fuca plate. Waveforms of 22 “shallow” earthquakes (15–30 km depth) and 16 “deep” earthquakes (40–60 km depth) were examined for shear-wave splitting. The fast-polarization direction is well-defined at 312±19° for the deep earthquakes, and 293±27° for the shallow earthquakes. These fast polarization directions are approximately parallel to the continental margin and coincide with the orientation of the pressure axes of small crustal earthquakes in this region. Interpreting this polarization pattern in terms of the extensive-dilatancy anisotropy model indicates a margin-parallel compressive stress regime. Comparing the average time delays between the fast and slow shear-waves for the deep earthquakes (0.32 s) and the shallow earthquakes (0.20 s) suggests that the source of the seismic anisotropy is stronger in the upper 20 km of the continental crust. Here, the average differential shear-wave anisotropy is estimated at 2.3%. These observations support the theory of weak coupling between the downgoing Juan de Fuca plate and the overlying North America plate, as the principal stress is perpendicular to the direction of subduction, and the S-wave splitting above the subducting plate is nearly perpendicular to the SKS splitting direction in the upper mantle beneath the subducting plate.

We investigate a possible isotropic component in the source of the Bolivian earthquake of 09 June 1994 through a study of the radial modes p S 0 (p = 0,1) excited by the event. Our study departs from previous similar attempts by making use of a time series of sufficient duration (over 116 days) to sample the shape of the spectral line of the fundamental radial mode 0 S 0 , and by combining measurements of two radial modes, thereby lifting the trade-off between deviatoric and isotropic sources of excitation. Our result (an isotropic component of 0.5 ± 1.3% of the main double-couple) fails to identify a significant change in volume in the source of this, the largest deep earthquake ever recorded.

The earthquake surface ruptures on the northern side of Awaji Island accompanying the 1995 Southern Hyogo Prefecture Earthquake in Japan consist of three earthquake surface rupture zones called the Nojima, Matsuho, and Kusumoto Earthquake Surface Rupture Zones. The Nojima Earthquake Surface Rupture Zone is - 18 km long and was formed from Awaji-cho at the northern end of Awaji Island to Ichinomiya-cho. It occurred along the pre-existing Nojima geological fault in the northern segment and as a new fault in the southern segment. The northern segment of the Nojima Earthquake Surface Rupture Zone is composed of some subparallel shear faults showing a right-step en echelon form and many extensional cracks showing a left-step en echelon form. The southern segment consists of some discontinuous surface ruptures which are concentrated in a narrow zone a few tens of meters in width. This surface rupture zone shows a general trend striking north 30°-60° east, and dipping 75°-85° east. The deformational topographies and striations on the fault plane generated during the co-seismic displacement show that the Nojima Earthquake Surface Rupture Zone is a right-lateral strike-slip fault with some reverse component. Displacements measured at many of the outcrops are generally 100-200 em horizontally and 50-100 em vertically in the northern segment and a few em to 20 em both horizontally and vertically in the southern segment. The largest displacements are 180 em horizontally, 130 em vertically, and 215 em in netslip measured at the Hirabayashi fault scarp. The Matsuho Earthquake Surface Rupture Zone striking north 40°-60° west was also found along the coastline trending northwest-southeast in Awaji-cho for ∼1 km at the northern end of Awaji Island. The Kusumoto Earthquake Surface Rupture Zone occurred along the pre-existing Kusumoto geological fault for ∼ 1.5 km near the northeastern coastline, generally striking north 35°-60° east, dipping 60°-70° west. From the morphological and geomorphological characteristics, the Nojima Earthquake Surface Rupture Zone can be divided into four segments which form a right-step en echelon formation. The geological and geomorphological evidence and the aftershock epicenter distributions show clearly that the distributions and geometry of these four segments are controlled by the pre-existing geological structures.

The sources for thirteen deep focus earthquakes ofm b≥5.5 andh>400 km in Northwest Pacific region were studied using waveform fitting and shear fracture source model. The source parameters were obtained as follows: focal depth, faulting plane, slip direction, rupture velocity, rupture length, rupture direction and scalar moment tensor. It was found that all these earthquake sources can be interpreted as shear faulting and have simple source time functions. The strike direction of faulting plane for most deep focus earthquakes coincides with that of the subduction zone, especially in the deep part of the subduction zone, it results in the tendency of reducing the dip angle of the subduction zone. The multiple point source model was also used to study the source process. The waveform fitting is better than the shear fracture model, but the general rupture direction which coincides with that from unilateral shear source model can not be obtained from the multiple point source model.

The effects of a strong non-linear thermo-kinetic coupling in the olivine-spinel transition are investigated with a 2D model for fast-subducting slabs. The thermo-kinetic coupling problem is solved in the Lagrangian framework by the differential algebraic method and by using experimental input parameters. Narrow regions with metastable olivine are found in fast and cold slabs down to the 660 km discontinuity, while the thermo-kinetic coupling reduces the depth and width of the zone where olivine and spinel coexist. In contrast to recent kinetic models, both latent heat release and heat diffusion place important constraints on the structure of the metastable wedge. Both processes trigger the ongoing phase transformation in the cold slab interior. A thermal instability over a short timescale (10 4 –10 5 yr) can occur due to a positive thermo-kinetic feedback. This effect is enhanced at greater depths. The depth of the metastable wedge is found to be extremely sensitive to certain geophysical and thermodynamic parameters such as the lithospheric basal temperature, slab width and the activation energy for diffusion. Variations in these parameters within the experimental uncertainties can cause significant changes in the depth of the metastable wedge. Hence predictions about the exact position of the metastable olivine wedge are rather difficult. We propose that deep focus earthquakes may occur due to a thermal runaway caused by shear heating, while the thermal instability is enhanced by the latent heat release associated with the olivine-spinel phase transition.

On June 8, 2008, at 12:25 GMT, a Mw 6.4 earthquake, the Movri Mountain earthquake, occurred in the area of northwest Peloponnese, western Greece. The epicenter was located in the municipality of Movri, 35 km southwest of Patras. For this study, a crustal anisotropy analysis was performed in the epicentral area of the Movri Mountain earthquake. Specifically, the shear-wave splitting phenomenon and its temporal evolution in relation to the Movri Mountain earthquake was studied, using the cross-correlation method. The data analysis revealed the presence of shear-wave splitting in the study area. Both before and after the Movri Mountain earthquake, the polarization directions of the fast component of the shear waves followed a general NNW-SSE direction. The observed mean fast polarization direction was not consistent with the regional stress field, which showed a general E-W direction of the maximum horizontal compressive stress. The differences between the estimated fast polarization directions and the properties of the regional stress field suggest the presence of a local stress field in the area around the fault. An increase in time delays was observed soon after the Movri Mountain earthquake. The average value of the delay times before the earthquake was ca. 18 ±2.6 ms, while after the earthquake this was ca. 40 ±4.6 ms. This increase in the time delay indicates changes in the crustal properties, which were possibly caused by variations in the preexisting micro-crack system characteristics related to the Movri Mountain earthquake, and the possible involvement of over-pressured fluids.

We have located and estimated source parameters of 109 earthquakes (0 to 5 M) using seismograms recorded at 2.5 km depth in the Cajon Pass borehole, southern California. The borehole is about 4 km from the San Andreas fault, at the boundary between the locked, almost aseismic Mojave and San Bernardino segments, where the San Jacinto fault approaches the San Andreas. This area is of interest both on account of its tectonic complexity and its high potential seismic hazard. The clear, relatively unattenuated downhole recordings are rotated to determine the incoming azimuth of the P wave, and the delay time between the P and S arrivals is used to estimate the hypocentral distance. The difference between the hypocenters located this way and the Southern California Seismic Network (SCSN) locations of the same earthquakes increases with hypocentral distance from the borehole. Of the earthquakes within 20 km of the borehole, 95% are within 1–2 km of the SCSN epicenters and all are within 5 km of the SCSN depths. All but three of the 58 earthquakes located here which the SCSN did not record were within 20 km of the borehole and so the location errors are similar to those of SCSN A and B quality events. Most of the small earthquakes occurred within the relatively active San Jacinto fault zone, but at least eight are located close to or within the San Andreas fault zone. The stress drops for these events range from 0.1 to 18 MPa. These earthquakes appear similar to the other events in this study suggesting that the San Andreas fault is similar to other faults at the scale of these small earthquakes. The variation in stress drop and slip orientation of these small earthquakes suggests that larger scale heterogeneity continues to these small scales. Also, the location of these small earthquakes in major established fault zones implies that earthquake source dimensions are not geometrically controlled by fault zone width.

Various types of energy can be stored in metastable materials at high pressure. If appropriate forms of energy released at a reasonable fast rate (⩽ 10 2 ) become available, some of the released energy must radiate elastic wave in a solid body. The energy efficiency in earthquake production for an earthquake having a magnitude of 7 or 8 was found to be 0.1%. “Earthquakes” caused by metastable phase transformations occurring under high pressure and temperature conditions have recently been confirmed in the laboratory by studies of the α → γ phase transformation in olivine. It is proposed that most earthquakes occurring at depths greater than ca 50 km inside the Earth represent the loci of the kinetic boundaries of the various phase transformations known to take place in the mantle silicates. Even the peculiar wedge-shaped double seismic zone observed between 70 and 180 km in the intermediate-focus earthquakes and the deep double seismic zone observed between 300 and 450 km can be interpreted in terms of this model.

According to the 3-D velocity images of the crust and upper mantle in Xinjiang and its adjacent areas and the projective distribution of epicenters at different depths, this paper reveals the relationship between the velocity structure feature of the crust and upper mantle and seismicity in the studied areas, and also reveals the deep seismogenic environment in the Tianshan earthquake belt, intermediate and deep focus earthquake region in Hindu Kush, intermediate earthquake belts (regions) in west Kunlun and Kalakorum Mountain and so on different earthquake belts. The seismogenesis is also discussed.

To explore a possible mechanism of deep earthquakes, this paper analyzes the unstable propagation of a stress-induced phase transition which is initiated in a homogeneous stress field. This Stephen problem is formulated as an initial-value problem for the phase boundary, and the driving force of the boundary is computed by using the solution of the boundary-value problem for a partially transformed material. The propagation of the phase transition under uniform pressure is numerically simulated. It is shown that (1) under lower pressure, the transition is terminated at a certain size, but it can propagate unstably when an initially transformed region is sufficiently large; and (2) when the pressure attains a critical value, the propagation becomes unstable, and goes in a particular direction depending on the initial shape. These results confirm the possibility of the unstable propagation of phase transition, and provide a theoretical basis for the hypothesis that the phase transition of a mantle material can trigger a deep earthquake.

A waveform inversion scheme is presented for multiple shocks. An earthquake source is modeled as a sequence of point sources with different focal mechanisms. Each point source is defined by an onset time, relative spatial coordinates, and a moment tensor. All the point sources are allowed to occur at only grid points with the same interval on a single flat plane and to have the same trapezoidal time history. Unknown source parameters are obtained by an iterative procedure. Each iteration step consists of two substeps; first, onset times and relative spatial coordinates of subevents with respect to a reference point are determined. Second, seismic moment-tensors of the subevents are obtained by a conventional linear moment-tensor inversion. A statistical test of a change in the focal mechanism during the source process is made using confidence ellipses of principal stress axes for each subevent. We apply the inversion to WWSSN teleseismic long-period P waveforms from a deep focus earthquake of the 1972 Banda Sea (mb = 6.6). The result suggests that the earthquake consisted of two major subevents with a time interval of 7.5 s, and that the main rupture, connecting the two subevents, propagated westward by approximately 20 km. The best double-couple seismic moment is 5.7 × 1018 Nm and 2.0 × 1018 Nm for the first and the second subevents, respectively. Differences between directions of the P- and T-axes of the two major subevents may suggest that the main rupture of the earthquake propagaged along a northward convex fault surface.

Radial modes excited by the 9 June 1994 Bolivia earthquake are analyzed to improve constraints on bulk attenuation in the Earth's interior. Using data recorded at 53 stations, quality factors for six radial modes are determined. The new observations are combined with previously published surface wave and normal mode attenuation measurements, and inverted for radial variations in shear and bulk attenuation, Qμ−1(r) and Qκ−1(r). The consequences of restricting finite Qκ to specific regions of the Earth's interior are investigated using a layered model parameterization. The optimal model is characterized by: 1) no bulk attenuation in the core, 2) a ratio of bulk to shear dissipation in the asthenosphere (∼30%) consistent with the presence of partial melt, and 3) a ratio in the remainder of the mantle (∼1%) consistent with solid-state mechanisms of attenuation.

The Mw 8.3 Bolivia earthquake occurred on June 9, 1994, at a depth of 636 km. This is the largest deep event in recorded history and ruptured a portion of the down-going Nazca slab unknown to have ruptured previously. We recorded the main shock and aftershocks on the BANJO and SEDA portable, broadband seismic arrays deployed in Bolivia during this event. Myers et al. (this issue) identified and located 36 aftershocks (M>2) for the 10-day period following the main shock. We use a grid search technique to determine focal mechanisms for 12 of these aftershocks ranging in magnitude from 2.7 to 5.3. We compare the observed P to SV and SH ratios to a series of synthetics that represent different fault plane orientations. We find consistent focal mechanisms with the T-axis roughly horizontal and oriented approximately east-west, and the P-axis predominantly vertical. The aftershock focal mechanisms indicate a rotation of the P-axis within the slab from down-dip compression prior to the main shock to a near-vertical direction afterwards. This observation is consistent with the release of shear stress on the near-horizontal rupture plane and the subsequent rotation of the maximum compressive stress to a fault -normal orientation.

Rupture during this event has two characteristics. First, bursts of major moment release (subevents) have a component of eastward movement from the epicenter. Second, later subevents show northward directivity. These observations lead to a model with four subevents, the latter three being line sources propagating within approximately ±30° of north. This model has en echelon regions of moment release over sub-horizontal plane(s) and rupture speeds up to 3.5±0.5 km/s within subevents. Moment release over en echelon regions can accommodate high moment release over a limited seismogenic volume within subducted lithosphere. Depending on the exact geometry of the Wadati-Benioff zone and the precise direction of rupture propagation, lower bounds on the thickness of the seismogenic zone during this event range approximately 10-40 km. While several aspects of the en echelon rupture model dovetail with predictions based on the anticrack model for deep-focus seismicity, one must also allow for the possibility that the rupture zone extended beyond the expected thickness of metastable olivine wedge.

The vertical ground displacement associated with the 9 June 1994 great Bolivia earthquake is calculated by summation of Earth's normal modes. Because of the great depth of the earthquake, the contributions to the surface displacement field from modes of low radial and angular orders predominate. Convergence is fast, and the truncated mode sum provides a good estimate of the total global displacement field. The vertical displacement is calculated to be on the order of 1 cm within a few hundred kill of the epicenter, and 1 mm at distances as great as 1500 km. Normal mode seismograms are calculated for a location near the epicenter, and the static offset is seen to occur primarily between the P and S phases, consistent with the effect of a near-field term. The predicted seismograms are in qualitative agreement with the recently reported observations of seismically recorded static displacements on the BANJO broadband array [Jiao et al., this issue]. The static offset calculated in this study is smaller than the observations and predictions reported by Jiao et al., probably owing to uncertainties in the observations and in their approximate calculations, which are based on dislocation theory in an elastic halfspace.

We have examined the source process of the June 9, 1994 Bolivia earthquake in the period range of body waves to free oscillations. A series of moment tensor inversions were performed on various teleseismic wave types and frequency bands to infer the source parameters and to map the temporal character of the source process. Time independent moment tensor inversion of the body waves (P and SH) yields a source time function of 46 seconds duration with four major “pulses” of moment release, with moment release dominated by the third pulse. The focal mechanism is dominantly dip-slip on the sub-horizontal nodal plane (strike 287°, dip 8° and slip −74°). The best bodywave fit requires a non-double-couple component of approximately 7%; although this could be due to an isotropic component of the source, it is at the margin of resolution. Time dependent moment tensor inversion of the body waves, which yields independent time histories for each moment tensor element, shows that the source orientations (fault plane and slip direction) change very little throughout the rupture duration. Analysis at several frequencies shows that the seismic moment is very stable (3.0 × 1021 Nm ± 6%) at periods greater than 150 seconds and that the isotropic moment is negligible given the resolution. We also modeled the P-waves for the January 10, 1994 event, the strong aftershock of June 9, 1994 and the August 8, 1994 earthquake. Rotation of the P axes for events after the mainshock may suggest a possible stress adjustment.

The June 9, 1994, Bolivia earthquake and the October 4, 1994, Kuril Islands earthquake provide a unique opportunity for a splitting analysis of the Earth's long-period toroidal free oscillations. We present toroidal splitting observations for the 6 fundamental modes 0T4, 0T5, 0T6, 0T7, 0T8, and 0T9, and for the 5 first overtones 1T1, 1T2, 1T3, 1T4 and 1T6. The first identification of mode 1T1 is made possible by simultaneously fitting the multiplets 1T1 and 0T7. The observed splitting is reasonably well explained by current shear velocity models of the mantle. Our analysis yields new estimates of the center frequencies and quality factors of these 11 toroidal modes.

The 1994 Bolivia and Tonga events show large differences in source and aftershock properties. The Bolivia earthquake, characterized by high stress drop, slow rupture propagation, and a weak aftershock sequence, appears similar to other large deep earthquakes that are isolated from active seismic zones, including the 1970 Colombia earthquake. In contrast, the 1994 Tonga event showed stress drop, rupture propagation velocities, and an aftershock sequence similar to that found for shallow earthquakes. The aftershock sequence of the 1994 Tonga event is many times stronger than has been observed for any other deep earthquake. A survey of deep earthquake aftershocks suggests that aftershock occurrence is correlated with the overall subduction zone magnitude-frequency relation (b-value). The largest deep events occur in isolated regions lacking smaller earthquakes, or in subduction zones characterized by anomalously low b-values. The 1994 Tonga event is an exception and is the largest deep earthquake to occur within an active subduction zone showing a large b-value, perhaps explaining the strong aftershock sequence. The Bolivia and Tonga events may represent end members of a population of deep earthquakes showing exceptional diversity in source properties, or they may represent two fundamentally different types of deep earthquakes, perhaps resulting from different physical mechanisms.

We investigate the physical setting of the Bolivian shock based on the history of the subducting Nazca plate, intraslab seismicity, deep seismic moment release, and seismic tomography. South America has two broad regions of reverse arc curvature. Subduction constrained to this unique geometry produces slab kinking contortions that may cause unusual slab thickening as they sink to the bottom of the transition zone and encounter resistance to penetration into the lower mantle. Such contortions are observed at intermediate depths “upstream” from the slab source regions of both the great 1970 Colombian and 1994 Bolivian events. Thickening helps explain how the Nazca slab accommodates large seismic source dimensions at depths of 625–650 km.

The Bolivia earthquake was remarkable for being felt in North America, as far as 8680 km from the epicenter. This paper examines the characteristics of ground motions in North America from the perspective of strong motion seismology. Data on ground motions is mostly from high-gain stations, but we also have two strong motion accelerograms. The largest accelerations and highest frequencies are associated with the P and PcP phases, which also correlate with the felt reports in several instances. Felt locations are associated with a combination of favorable factors. Most are in a distance range where P-wave amplitudes tend to be relatively high, as observed in the empirical correction factors for body wave magnitude, and in regions with relatively high values of Q in the lithosphere. In addition, site effects, structural amplification, or both, seem to contribute to the selective detection of the earthquake in the regions where some people noticed it. Motions were far below design spectra for earthquake resistant design, and far below levels likely to be associated with damage.

Case studies of four multiplets (υs-sensitive 1S7 and 1S8 and υp-sensitive 5S5 and 5S6) are considered to test the accuracy of normal mode constraints on aspherical structure at degrees above 2 (4,6,8). Analyses of misfit, along-branch consistency, and consistency with existing mantle models argue for the feasibility of constraints on higher structural degrees. Preliminary interpretation indicates a relative insensitivity of data misfits to dlnυp/dlnυs at degrees 6 and 8 in the lower mantle (perhaps due to inaccuracies in current δυs models), and provides weak evidence of some decorrelation between δυs and δυp in the middle and lower mantle.

The source time function of the great 1994 Bolivia deep earthquake is estimated from far-field seismograms by two techniques : time-domain inversion of direct P and S waveforms, and spectral-domain inversion of a combination of body waves, first-orbit traveling modes, and free oscillations. Both methods give consistent estimates, but the latter provides more bandwidth (1-300 mHz) and is more robust with respect to source-directivity effects and wave scattering. The earthquake released a total moment of 2.7 (± 0.1) x 10 21 Nm over an interval of about 50 s. Its history is highly episodic and can be divided into three stages : an initial sequence (0 ≤ t ≤ 13 s ; 2 x 10 20 Nm), a main-event sequence (13 ≤ t ≤ 38 s ; 2.2 x 10 21 Nm), and a terminal sequence (38 ≤ t ≤ 50 s ; 3 x 10 20 Nm). The main event is resolved into five subevents with peaks at - 16, 20, 26, 29, and 36 s. The spatial pattern of moment release suggests that the main event was confined to the cold slab core, while the initial and terminal stages were part of a distinct westward-propagating rupture in the lower reaches of the former oceanic plate.