Using near-field, broadband data from a suite of crustal earthquakes which occurred in a small, homogeneous source region, we examine the scaling relationships among source parameters for earthquakes in the critical 1.5 to 4.5 magnitude range. We take special precautions to insure that our conclusions are not biased by local structure. Seismic moment, fault radius, and stress drop relations were determined for 32 small, shallow, strike-slip earthquakes (1.6 ≦ ML ≦ 4.3) in the Matsushiro region of southwest Honshu, Japan.
Spectral and cepstral analyses of short-period arrivals exhibit structure interpreted as resulting from reverberations within thin, low-velocity surface layers below the recording site. Furthermore, calculated crustal transfer functions for SH predict a peak over the range of 2 to 4 Hz. This type of site response can bias estimates of spectral parameters; particularly corner or characteristic frequencies. For simple sources, such a bias can be reduced by filtering in the cepstral domain. The effect of cepstral filtering on corner frequency determinations is generally less than 20 per cent and tends to decrease them, particularly for larger events where the structurally induced spectral enhancement is at frequencies above the source-related corner frequencies.
Estimates of the seismic moment, Brune stress drop, and fault radius for each event were calculated from SH displacement amplitude spectra. The seismic moments range from 1018 to 1022 dyne-cm. The stress drops range from 0.008 to 16 bars. The relationship between log-stress drop and log-moment is approximately linear with a slope of 1 up to a moment of 2 × 1021 dyne-cm, but for higher moments the slope is not significantly different from zero. Fault radii range from 0.3 to 1.2 km. There is a physical scatter in radius for any given moment up to a factor of 1.5. For moments less than 2 × 1021 dyne-cm, there is no significant trend in the radius as a function of moment, but for higher moments the radius increases with moment.
The main conclusion is that the stress drop is the dominant scaling factor for moments less than 1021 dyne-cm. For higher moment, the source dimension becomes the controlling factor.