SH ground-displacement spectra (1 to 12 Hz) for 16 local earthquakes (Δ ≦ 18 km, 1.1 ≦ M ≦ 4.6) recorded at a common site situated atop the active trace of the San Andreas fault are used to estimate attenuation characteristics for propagation paths along the fault trace. t* = 0.10−0.13 (corresponding to an equivalent total path Qβ = 75−100) is appropriate for events with focal depths of ∼ 10 km.
Propagation-path effects, and not processes at the earthquake source, control corner frequencies for small (M ≲ 3) earthquakes for these highly attenuating paths. The results obtained here suggest that as a rule of thumb, if the true equivalent total path Q is as low as 4·ƒc·t, where ƒc is the estimated corner frequency and t the travel time, the corner frequency estimate is determined by propagation-path effects, not by processes at the earthquake source. In these cases, reliable estimates of source parameters can only be obtained if the appropriate propagation-path corrections are known.
Using Brune's model of shear-wave spectra, source dimensions L = 2r of less than 250 meters and stress drops greater than about bar are estimated for the smaller events (1.1 ≦ M ≦ 2.2), using the equivalent total path Qβ obtained here. The seismic moments obtained in this study, together with data for larger central California events (2.4 ≦ ML ≦ 5.1) obtained by Johnson and McEvilly (1974), imply a linear log seismic moment-magnitude relation for 1 < ML < 5