The J-Array is a large-aperture seismic array system in Japan, which consists of 14 regional seismic networks deployed for research on earthquake prediction. During the project period (1991-1992), many seismograms from 61 teleseismic events (M.≥5.8) were collected for studies on structures deep inside the Earth. A report summarizing J-Array seismograms was published in 1994 in order to show the characteristics of seismograms and evaluate the quality of their waveforms. In this paper, the characteristics of J-Array seismograms are introduced and discussed briefly. The seismograms from a large, dense distribution of seismic stations have a high potential to detect weak onsets of unknown phases, which may reveal hidden structures in the Earth's interior. Using the J-Array seismograms, two distinct unknown phases are presented and their tentative interpretations are proposed. One of them is a reflection from the 410 km discontinuity in the upper mantle, and the other is from an unknown discontinuity in the lower mantle with a depth of about 1, 200 km. Both phases might have been missed, if the J-Array did not have a high potential to detect weak and obscure seismic phases, thereby proving the superiority of the J-Array. The limitations of the J-Array also became clear. It is not an ordinary seismic array in a strict sense, but joint observations of regional seismic networks. The use of non-uniform sensors and telemetering systems reduce the resemblance of waveforms among stations, and weaken the ability of the J-Array. In order to enhance the potential of the J-Array, these problems must be solved. The "New J-Array" project, which will start in 1995, is expected to resolve such problems.

View full abstract

A major field-based observation program (the "Skippy Project") is underway to synthesize an array covering the whole Australian continent, using a series of deployments of around 10 broadband seismometers for periods of 5 months at a time. Such a time period is long enough to exploit the favourable position of Australia relative to world seismicity to assemble a good sampling of global structure. The distance coverage from available events is very good, so that it is possible to build up composite record sections covering propagation from 10 to nearly 180.., with a full coverage of the body and surface wavefields. For deep events at regional ranges, the field arrays also yield good data on mantle reverberation patterns.

View full abstract

We introduce the waveform station anomaly (WSA) of teleseismic P-waves to investigate the crustal and uppermost mantle structure beneath stations, which is a natural extension of relative travel-time and amplitude anomalies. WSA is defined as the observed waveform deconvolved by the average waveform of stations in an array. Since source time function and mantle response are almost identical across the array and cancelled out by the deconvolution, WSA approximately represents the effect of the receiver structure and the instruments peculiar to each station. The procedure was applied to short-period teleseismic waveforms recorded by the J-Array, a large-aperture and short-period seismic array covering the entire Japanese Islands. WSA was consistently determined for a cluster of deep events in the Tonga-Fiji region, with almost identical back azimuth and incidence angle. Especially, near-station reverberations inherent to each station were effectively extracted by the deconvolution. However, relative travel-time residuals showed a slight scatter due to the unequal number of stations used for analysis of each event. Despite the small number of events used in this study, WSA clearly changes with the arrival direction, which indicates that we have to consider a laterally heterogeneous structure beneath the Japanese Islands to explain the observed WSA.

View full abstract

An analysis of hundreds of seismograms from a seismic array (J-Array) in the Japanese archipelago shows travel-time anomalies on the order of 4-10 s for P-waves from teleseismic events in some source-receiver geometries. The source-receiver pairs exhibiting such large anomalies are 1) Philippine to Indonesia-Kanto, Japan, 2) Kamchatka-Southern Hokkaido to Tohoku, Japan, and 3) Taiwan to Ryukyu-Shikoku to Kinki, Japan. A simple ray-path consideration indicates that the rays responsible for these anomalies are those incident to the base of the Izu-Bonin slab from below at near critical angles and those propagating laterally within the Kurile slab and the Ryukyu slab along their strikes. The travel times of the first arrivals are calculated for slab structural models using the method of Sakai (1992), which traces progressive expansion of a wavefront away from the source based on the Huygens' principle. This method has an advantage over conventional ray tracing methods in its robustness against the complexity of a laterally heterogeneous velocity structure. A slab model is specified by its thickness and velocity contrast with the surrounding mantle. Numerical experiments show that the effect of slab thickness is different from the effect of velocity contrast, and that their effects are separable. With an important assumption that the IASP91 model approximates the surrounding mantle, the observed negative anomalies can be consistently explained by a 50-km thick slab with a velocity contrast of 5% for the Pacific plate and by a 30-km thick slab with a velocity contrast of 5% for the Philippine Sea plate, although the observed records exhibit negative anomalies that are unable to be explained by the mere introduction of a high-velocity slab.

View full abstract

Broadband seismic waveform data recorded at stations in the western Pacific region are analyzed to investigate mantle discontinuities. By using teleseismic deep events, we observe unambiguous P-to-S conversion waves associated with the mid-mantle discontinuities in many of the individual seismograms. The commonly used method of stacking receiver functions is not so effective in this case due to the limited number of deep events. An inversion scheme is developed to determine a discontinuity response function at each station by fitting observed waveform data with superpositions of the P-to-S converted waves. Beneath the station in northeast China (MDJ) where the subducted Pacific plate appears to stagnate along the "660-km" discontinuity, the discontinuity response function has more complicated features than those of other stations. The preliminary results indicate no depression of the "660-km" discontinuity at the tip of the subducting slab beneath MDJ; instead a multiple-discontinuity structure down to a depth of 780 km is observed.

View full abstract

The upper-mantle P-wave velocity structure beneath the northern Japan arc is investigated based on the data recorded in the J-Array. The data set includes 260 seismograms and 1, 817 travel time picks including 274 later arrivals from 33 events in the Kuril-Kamchatka region. Seismograms are slant-stacked and iteratively downward continued to obtain the upper-mantle structure, which well explains the observed data. The travel times and relative amplitudes of later arrivals are used to give additional fine constraints on the structure around upper-mantle discontinuities. The most prominent feature of the estimated model is the 410 km discontinuity, which is deeper (420 km) and has a larger velocity jump (around 8.0%) than those in previous studies for neighboring regions. This velocity jump and a transition width less than 10 km are consistent with the calculated or experimented properties of the equilibrium phase change of olivine. The depth of 660 km and 5.6% velocity jump at the 660 km discontinuity are less reliable because corresponding later arrivals are not picked. The 240 km width of the transition zone is not reconciled with the model of cold and olivine-rich mantle beneath subduction zones. The model has additional minor inflections at depths near 540 and 780 km. Another major characteristic of the model is the lack of a low-velocity zone. In the uppermost mantle, the velocity increases linearly at depths between 35 and 85 km and is almost constant between 85 and 130 km.

View full abstract

We identified diffracted P-waves, P_diff, in 15 short-period records from Global Digital Seismic Network data, to infer heterogeneities near the core-mantle boundary (CMB). These records are classified into three types by their waveforms: 1) a clear phase observed at the predicted arrival time of Pdiff with late arrivals of minor energy; 2) continuous high-energy arrivals, called tail, lasting for 30-50 s after an initial phase; and 3) an initial phase observed unclear and a long tail with large amplitude of more than 50 s. From particle motions in the tail and comparisons between Pdiff and direct P-waves propagating through the lower mantle, the characteristics of each type are caused by scattering due to heterogeneities deep in the Earth, probably near the CMB. Assuming that the corresponding heterogeneities are only the result of irregular core-mantle topography, we estimate the CMB configuration from the above observations and some simple numerical simulations including the multiplescattering effect. The first type of Pdiff corresponds to a core-mantle topography with a small number of bumps, while many bumps of various sizes exist for the other two types. We cannot explain the long duration and high-energy level in the tail of the third type only by the scattering from CMB irregularity. The Iong tail of the third type suggests the existence of some structures that trap scattered waves effectively, such as a layer where the P-wave velocity gradient is negative or a low-velocity zone just above the CMB.

View full abstract

The attenuation of P-waves and its frequency dependence in the Earth's outer core are investigated using the P4KP/PcP spectral amplitude ratio. We analyze PcP and P4KP phases recorded on short-period seismographs in the Japan Islands for three large intermediate-depth and deep earthquakes from March 1991 through July 1994. The observed P4KP/PcP spectral amplitude ratio is corrected for responses of the mantle and the core, and attenuation in the lower mantle. The responses for P4KP and PcP are considered for the frequency dependence of their reflection/transmission coefficients at the core-mantle boundary. After the corrections, we estimate Q_P, values in the outer core in the frequency range of 0.15-2 Hz, and find that the Q_P increases with increasing frequency (f) in the form of Q_P=(4, 600±100) f(^0.55±0.07). The observed frequency-dependent Q_P values increase more slowly with increasing frequency than those predicted by the model of the bulk attenuation based on the relaxation mechanism of immiscible particles in the outer core.

View full abstract