Effects of lateral velocity heterogeneity under the Nevada Test Site on short-periodP wave amplitudes and travel times

Short-period teleseismicP waves from the Nevada Test Site (NTS) show systematic variations in amplitudes and travel times, with low amplitudes corresponding to fast travel times, suggesting elastic focussing-defocussing effects. Also, the azimuthal amplitude and travel time patterns for events at the Pahute Mesa subsite are systematically different from those at the Yucca Flat subsite, indicating the presence of a near-source component in both the amplitude and travel-time variations. This component is isolated by removing the mean station pattern for all of NTS from the observations. A very-near-source component in the Pahute Mesa observations is also isolated by removing subsite station means from the measurements, whereas the Yucca Flat observations exhibited no coherent very-near-source component. These anomalies are back-projected through laterally homogeneous structure to form thin lens models at various depths. Travel-time delays are predicted from the amplitude variations using the equation for wavefront curvature. The long-wavelength components of the predicted and observed time delays correlate well, at depths of 25 km for the very-near-source component under Pahute Mesa and 160 km for the regional component under NTS. The time delay surfaces predicted by the amplitudes at these depths are mapped into warped velocity discontinuities suitable for the calculation of synthetic seismograms using the Kirchhoff-Helmholtz integral formulation. Both the intersite (near-source) and intrasite (very-near-source) differences in amplitudes are qualitatively predicted very well, although the range of variation is somewhat underpredicted. This deficiency is likely due to the destructive interference of anomalies inherent in back-projection to a single layer.