Preliminary analysis of resolving power of existing strong-motion arrays for source inversion.

Abstract

The effectiveness of existing strong-motion arrays for source inversion analysis is investigated by estimating the accuracy of the inversion solution using simplified theoretical seismograms. The resolving power is examined for different subfault sizes and for different array configurations for four existing array networks. They are for the 1979 Imperial Valley and the anticipated Parkfield, California earthquakes, and the 1968 Tokachi-Oki and the anticipated Tokai, Japan earthquakes. We use the method previously developed by Miyatake, Iida, and Shimazaki in 1986 based on the Wolberg's prediction analysis. The main results are: (1) The array used for an analysis of the 1979 Imperial Valley earthquake is not suitable for source inversion; especially the El Centro array, a linear array crossing perpendicularly the fault trace, is ineffective. (2) The array installed for the anticipated Parkfield earthquake by the end of 1985 seems to be satisfactory because of the intensive installation of many stations. However, the resolving power of the whole array will further increase by adding a few stations in the northwestern part of the fault or in distant areas for a better azimuthal coverage in the northwest of the fault. (3) Detailed source inversion analysis cannot be expected for the 1968 Tokachi-Oki, Japan, earthquake because of both the large fault area and lack of offshore stations. Strong-motion, ocean bottom instruments within the fault area are required for a further improvement of the inversion analysis for this type of earthquake. (4) An addition of several land stations on the west and north sides of the fault area is desirable to the present network for the anticipated Tokai, Japan earthquake. The resolving power of the whole array for this earthquake is strongly dependent on a rupturing direction because station coverage of the southern oceanic part of the fault tends to be poor.