THE INFLUENCE OF CRUST AND UPPER-MANTLE HETEROGENEITY ON SHORT-PERIODWAVE-FORM DISTORTION

Short-period teleseismic P waves may suffer considerable distortion ow ing to the effects of upper-mantle/crustal heterogeneity. This sensiti vity suggests the use of P waveforms as an aid to constraining upper-m antle structure; however, this requires a means of separating the rela tive contributions from these two regimes. We investigate the nature o f short-period waveform distortion as recorded on the Washington Regio nal Seismograph Network (WRSN). A technique for analyzing short-period teleseismic P-waveform distortion is presented to extract the impulse response function of the underlying crust and upper-mantle for a give n event-station pair. The technique is shown to yield highly reproduci ble results and permits direct comparison of response functions for di fferent source locations. Short-period waveform distortion can be stro ng and is essentially incoherent between stations (minimum separation 20 km). We subsequently confine our attention to a single station wher e waveform distortion is evidenced by strong defocusing and examine th e behavior of this distortion as a function of source location. In add ition to crustal defocusing, which persists over a broad range of azim uths and epicentral distances, we also find evidence for more subtle v ariations in the initial waveform which may be attributable to upper-m antle structure. These observations are modeled in two ways. We employ a parabolic approximation method to simulate distortion by deep-seate d, large-scale structure as exemplified by a best estimate model of th e upper mantle below WRSN derived from travel-time inversion. The resu lting synthetics fail to exhibit strong waveform distortion; the princ ipal effects are focusing/defocusing which accompany geometrical sprea ding. Crustal heterogeneity is probably stronger and exists at smaller scales comparable to the wavelengths of the short-period waves, hence contributions to waveform distortion through scattering processes bec ome important. These contributions are investigated qualitatively usin g an acoustic finite difference code and models of crustal heterogenei ty comprising volume heterogeneity and interface topography. Waveforms vary markedly over station separations as small as 5-10 km but at a g iven station are relatively insensitive to changes in source location. Crustal contamination will seriously complicate the retrieval of info rmation on upper-mantle structure from short-period waveform distortio n; however, these results suggest that the identification of an upper- mantle signature can be achieved by comprehensive analysis of the impu lse response at a given station as function of epicentral distance and back-azimuth.