Abstract:
The principal difficulty in finding detailed tomographic images is caused by signal-generated noise such as that due to adjacent inhomogeneities. This difficulty cannot be overcome by any of the techniques available in seismic tomography. The degree of detail attainable in tomographic images can be considerably enhanced by the method we put forward in this paper, namely, successive subtraction of velocity anomalies (SSA method). The basic idea of the method consists in successive subtraction of time anomalies due to the parameter that is the most responsible for increasing the sum of squares of the residuals. In mathematical terms this means that the inversion relies on a technique from linear algebra known as a coordinate method of possible directions. This algorithm was applied to 3181 P-wave travel times recorded at seismic stations from 242 shallow and deep earthquakes in order to derive tomographic images around the epicentral area of the 1996 magnitude 5.9 Onikobe earthquake, northern Honshu, Japan. A resolution test showed that reliable solutions were for shallow blocks at locations of highest ray intersection density. The result corroborated that the method was able to clearly identify strong inhomogeneities and detect low contrast inhomogeneities near them. It was found that the epicenter of the 1996 Onikobe earthquake, its aftershock area, and the zone of highest seismicity rate, were all located close to contacts between positive and negative velocity anomalies.