P-wave spectral amplitudes from records of the Guerrero accelerograph array frequently equal of exceed the S-wave spectra at high frequencies (f > 10 Hz). Since the instruments are located on granitic or metamorphic outcrops with expected high values of Q, uncertainties due to attenuation are minimized. A range of plausible values of QP/QS established simple diagnostic tests to distinguish the path effects between P and S waves from those caused by the source. We also used wave arrivals that have traveled along similar paths and have been recorded on the same sites to eliminate the path effect.
It is likely that both attenuation and source processes contribute to the high P-to S spectra observed. Attenuation alone cannot account for such high ratios, and simple shear dislocation models do not predict them either.
To explain the results, Haskell's (1964) formulation was used to calculate the energy radiated by a shear fault, with energy from tensile motion superimposed. It was found that small motions normal to the fault can account for P / S spectral ratios close to one, suggesting that fault tensile motions may be causing the anomalous P-wave radiation observed. To be consistent with the entire data set, we propose that such tensile motions are a common but not universal feature of fault rupture.