Seismic intensity data, which measure the degree of ground shaking, contain information on attenuation along the path from the source to the station and earthquake source strength. With some reasonable assumptions about the seismic intensity data, a properly formulated damped least squares estimation procedure can be used to determine simultaneously both three-dimensional attenuation structure and source strength. By means of this method, sufficiently distributed earthquakes with a large amount of seismic intensity data may provide information on the crust and upper mantle Q structure in a region where a spatial variation of attenuation is large enough to affect the distribution of seismic intensity. To check the validity of the method, numerical experiments were undertaken for artificial data including errors comparable to those expected from using seismic intensity. The results obtained for a two-dimensional island arc structure model show the method to be practical. The reliability of the obtained solution for Q structure can be measured by the corresponding diagonal element of the resolution matrix. By using a large number of seismic intensity data up to about one thousand and selecting solutions with high resolution, a Q structure with an accuracy of 1.5×10^-3 in 1/Q and a source strength with an accuracy of 0.1 in magnitude scale can be obtained. This method is extendable to many regions including island arcs and continents because large amounts of intensity data are easily attainable.

View full abstract

For predicting strong ground motions from major earthquakes and for close investigations into complex rupture processes, two different analytical and semi-empirical approaches are used to synthesize seismic waves from a nearby fault with a large linear dimension. The former technique is to calculate Green's functions for a horizontally layered structure by the discrete wave-number/finite element method, and the latter is to use the records of minor shocks as empirical Green's functions by convolving a correction function for the differences in the source functions and the receiver responses between the main and smaller events. In both cases, the phase-delayed Green's functions are integrated over the entire fault surface.
The above methods have been applied to the case of the 1969 central Gifu earthquake (M=6.6) which was followed by moderate aftershocks (M=4.3-4.8) and a number of smaller events. It was found that the waveforms synthesized from the two approaches agree reasonably well with each other. The strong-motion records, particularly of body waves and the major portion of surface waves with periods longer than 5-7 s, can be satisfactorily modeled by the theoretical synthesis with a realistic structure and also by the semi-empirical analysis using four aftershock records, if reasonable rupture velocities and rise times are assumed.
However, the shorter-period waves with periods 1-2 s involved in the records cannot be simulated by either of these syntheses, unless incoherent rupture propagation over the fault is included. A stochastic fault model with variable rupture velocities over large-scale fault segments is tentatively presented to account for the short-period waves.

View full abstract

Upper mantle velocity structure in the New Guinea-New Britain-Solomon Islands regions of the Southwest Pacific has been studied to a depth of about 550 km from the analysis of P and S wave travel times data of 128 mantle earthquakes. Wave velocities were obtained at the depths of foci of earthquakes in the inclined seismic zones in these regions, by using KAILA'S (1969) analytical method. By a linear fitting of velocity variation with depth, it is found that the P velocity in the New Guinea region increases from 7.92 km/s at 40 km depth to 8.26 km/s at 230 km depth. On the other hand, the P velocity function for the Solomon Islands region reveals a velocity of 7.80 km/s at 40 km depth which increases, with a velocity gradient of 0.280±0.003 km/s per 100 km, to 8.81 km/s at a depth of 400 km. At this transition depth of 400 km, there is a sharp first-order velocity discontinuity-the velocity increasing from 8.81 to 9.44 km/s which is about a 7.2 % velocity increase across this discontinuity. Below this disco tinuity at 400 km, P velocity increases again from 9.44 km/s, with a gradient of 0.230±0.035 km/s per 100 km, to 9.77 km/s at a depth of 545 km. There is no evidence, from the present study, for the presence of a significant low-velocity layer in the inclined seismic zones beneath the New Guinea-New Britain-Solomon Islands regions. The velocities obtained in the Solomon Islands region are 3 to 6 % (on an average) higher and lower, respectively, than those in the Japan and the Tonga-Kermadec regions.

View full abstract

The stress-field near an arc-arc junction shows an anomalous state against the expectation from the relative motion of converging plates. The Hokkaido corner is a typical example of such anomalous regions, where the compressional axes (P-axes) derived from focal mechanism of earthquakes in the continental lithosphere lie parallel to the strike of trench axes.
In order to explain this anomalous state of stress, a three-dimensional finite element method is employed to model the configuration of the Pacific plate subducting beneath the Hokkaido corner. Several types of plate driving forces are taken into consideration: (1) a negative buoyancy due to the density contrast between the subducted lithosphere and the surrounding asthenosphere, (2) a negative buoyancy which might act on the lid of the Sea of Japan to simulate a presumed subduction of the "Japan Sea plate, " (3) a northwestward compressive force generated by the movement of the Pacific plate, (4) a northwestward compressive force due to the flow in the asthenosphere, (5) an eastward compressive force expected from the movement of Eurasia or the assumed "Japan Sea plate, " (6) a drag force exerted at the edge of the subducted slab in the model space, (7) thermal stresses arising in the subducted slab which are caused by the contact with a hot asthenosphere.
We calculate the displacements and stresses in the crust and upper mantle around this region, due to these forces under the assumption of viscoelastic rheology, compare them with the observations, and evaluate their relative importance.
It is concluded that a negative buoyancy exerted on the descending slab could yield the observed anomalous stress field and that all of the compressive forces under consideration might not play an important role. However, downwarping of crustal vertical movements is predominant over the whole region under consideration in the case of a negative buoyancy acting on the subducted Pacific slab. Other sources such as thermal stresses might be required, which yield upward motions and preserve the trend of P-axes. An anomalous wedge zone which might exist above the subducted slab on the landward side of the junction would have some effects on the generation of the complex stress field.

View full abstract