Dispersion patterns of the ground roll in eastern Saudi Arabia

Abstract

Seismic surveys on land must be designed so that the source‐generated noise, such as ground roll, is preferentially attenuated before P‐wave signal amplification and recording. The correct specification of spatial and frequency filters requires prior knowledge of the noise properties in the area. We show that the strong Rayleigh wave component of source‐generated noise has a wavelength range which is predictable on a regional scale, using widespread P‐wave velocity measurements in shallow upholes. This predictive capability decreases the number of noise analyses required to map the boundaries between areas with different Rayleigh wave properties. The case history presented is for northeastern Saudi Arabia, an area of roughly [Formula: see text]. The data comprise 80 noise analyses and a data base of over 10,000 up‐hole measurements of P‐wave velocities, supplemented by maps of topography and geologic outcrops. Examples show that the frequency‐wavenumber transforms of time‐offset records can be interpreted in detail in terms of Rayleigh wave dispersion and air wave coupling, dictated by the elastic properties of the very shallow layers. P‐wave velocities, measured in shallow upholes at noise analysis sites, are used to form initial estimates of the corresponding shear‐wave velocities and subsequently refined by matching the observed and predicted dispersion curves. Even without this refinement process, the initial S‐wave velocities can be used to estimate Rayleigh wave velocities at frequencies which typify the top and bottom of current vibrator sweeps (10 and 80 Hz). These velocities are mapped for the area and used to determine the wavelength range of Rayleigh waves. An effort is also made to map regions where Rayleigh wave scattering from surface topography is likely to occur.