Fine structure of P_nl waves from explosions

Abstract

The long‐period character of P nl waves from many tectonic regimes has been modeled in the past using a simple homogeneous layer over a half‐space crust model. Scant attention has been paid to the short‐period wave form character. In this study the P nl waves from Nevada Test Site explosions as recorded on a regional digital seismic array have been modeled in the period range 1–10 s. The fine structure of the P nl waveform has been carefully analyzed. The effects of a realistic velocity gradient at the free surface or a velocity transition zone at the crust‐mantle boundary have been calculated. A free surface gradient of even a moderate nature has a major effect on the waveforms, whereas a finite crust‐mantle transition apparently does not. We have developed an efficient scheme to implement the wavenumber integration theory for computing short‐period P nl waves at regional distances. We utilized both the wavenumber integration and a generalized ray theory synthetic seismogram code for use in our analysis. The former was used to compute exact, though band‐limited, synthetics. The latter was used to determine which generalized ray paths contributed the primary energy to the major features of the waveforms. For the simple one‐layer crust model the dominant energy to the critically reflected P wave portion of the wave comes from the primary ray reflected from the mantle. We refer to this as.PmP similar to the notation of PcP for the primary core reflection. The next high‐amplitude energy is associated with a group of ray paths with at least one converted mode (i.e., P converting to S and vice versa). If the wave velocities are gradientlike within the top few kilometers of the model (6 km in this study), the contribution from 2PmP (a ray reflecting twice from the crust‐mantle interface) ray path becomes significant. The converted group of rays still contributes but not strongly. Free surface reflections become significant with the surface gradient present, whereas they are not important for the simple one‐layer model. Besides the direct Pn, four major ray groups have been specifically identified in the observed waveforms and shown to dominate the wave shape. These are the PmP and 2Pmp groups, the converted group, and the 2Pn head wave group. The 2PmP and 2Pn groups are portions of the same generalized rays. Since the converted group of rays loses significance in the presence of a realistic free surface gradient, the onset of the critically reflected portion of the Pnl waveform can be modeled as primarily an interaction between PmP and 2Pmp. The increased importance of depth phases in this part of the waves implies that it may be useful in improving source discrimination capabilities. Low‐Q surface layers have been explored as a possible physical mechanism for the resolving power of regional spectral discriminants. Such a mechanism appears plausible. If this is the dominant mechanism in differentiating earthquake and explosion spectra, then earthquakes should have a much more energetic PmP phase, with later arrivals be...

Intermediate depth earthquakes beneath the India‐Tibet Collision Zone

“…:ftectivity code developed by Saikia and Burdick (1991). The velocity model consists of two layers over a half space with the Moho at 64 km (Table 2).…”

Calibration of the Regional Crustal Waveguide and the Retrieval of Source Parameters Using Waveform Modeling

“…Generally this is done by modifying the overall thickness of the crust and velocities of the crustal layers and the upper-most mantle. The primary upper layer, corresponding to the sedimentary structure, strongly aects the Rayleigh waves and the character of the PL (extended P-wave) and S wavetrains (SONG et al, 1996;SAIKIA, 1994;SAIKIA and BURDICK, 1991). If observations suggest a strong gradient in the velocity structure ± that is, the initial P wave is sharp and impulsive with the character of turning ray rather than the longer period ramp-like head-wave character ± a detailed model with a gradient in the mantle structure must be used (SAIKIA, 1994).…”

“…LEFEVRE and HELMBERGER (1989) and HOLT and WALLACE (1989) demonstrated that the P nl analysis procedures can be applied without substantial modi®cation to observations from virtually any tectonic region. In a later study SAIKIA and BURDICK (1991) extended the application to shorter period P nl waves and demonstrated that short-period P nl waves also have many stable features which can be deterministically modeled to characterize the regional crustal waveguide. This characteristic of P nl waves is due to their primarily sampling the lower crust and the Moho interface, and consequently being relatively insensitive to lateral crustal variations, which occur mainly in the upper crust.…”

National Earthquake Hazards Reduction Program, summaries of technical reports; Volume XXXV

“…We computed synthetic seismograms for a magnitude 7.3 strike-slip earthquake using the synthetic seismogram method described by . We used a fast algorithm for computing Green's functions by frequency-wavenumber integration (Saikia and Burdick, 1991). The ground motions at a site 1.5 km from the fault rupture surface were estimated by averaging the ground motions calculated for a range of different rupture scenarios.…”