Separating the scattered wavefield from teleseismic P using curvelets on the long beach array data set

Abstract

A dense seismic array, composed of over 5000 stations with an average spacing close to 120 m was deployed in Long Beach, CA, by NodalSeismic and Signal Hill Petroleum as part of a survey associated with the Long Beach oilfield. Among many interesting wave propagation effects that have been reported by others, we observe that the coda of teleseismic P waves display waves caused by obvious local scattering from the Signal Hill popup structure between strands of the Newport-Inglewood fault. The density of the seismic array allows space-based methods, such as the Curvelet transform, to be investigated to separate the teleseismic and local scattered wavefields. We decompose a synthetic wavefield composed of a teleseismic plane wave and local scattered spherical wave in the curvelet domain to test the plausibility of our curvelet analysis and then apply the technique to the Long Beach array data set. Background noise is removed by a soft thresholding method and a declustering technique is applied to separate the teleseismic and local scattered wavefield in the curvelet domain. Decomposed results illustrate that the signal-to-noise ratio of the teleseismic P wave can be significantly improved by curvelet analysis. The scattered wavefield is composed of locally propagating Rayleigh waves from the pop-up structure and from the Newport Inglewood fault itself. Observing the wavefield both in space and time clearly improves understanding of wave propagation complexities due to structural heterogeneity.