Seismic Interferometry Applied to Seismic Background-Noise Field Data

Abstract

One of the applications of Seismic Interferometry (SI) is the reconstruction of the Earth’s reflection response from the crosscorrelation of seismic background noise recorded at the surface. In recent years, several authors have derived the relations that govern this process. The quality of the reconstruction has been extensively examined with numerical modeling results. Several authors have applied the method to real data for the reconstruction of surface waves. We applied SI to background-noise field data recorded in a desert area. The reconstructed results show several coherent events – inclined and nearly horizontal. The inclined coherent events are interpreted as reconstructed surface waves. The nearly horizontal coherent events appear to align well with reflections from an active survey along the same line. Therefore, we interpret these events as reconstructed reflections. INTRODUCTION If one looks at a raw seismic reflection (exploration) record, one notices that apart from the desired reflections and (mostly undesired) surface waves, there are a lot of other events that are normally described as ”seismic background noise”. And as a result, one of the goals of a data processor is to eliminate this unwanted noise. But there is also useful information in the seismic noise – it may contain propagating waves (transmission response) even though these are not a result of conventional exploration sources. Claerbout (1968) was the first to show how to make use of the seismic noise. He proved that for a 1D acoustic medium the autocorrelation of the transmission response can reconstruct the reflection response of the medium. Later, he stated the conjecture that in the case of a 3D acoustic medium, one should crosscorrelate the observed at the surface transmission response to reconstruct the reflection response. Wapenaar et al. (2002) proved Claerbout’s conjecture for any 3D inhomogeneous acoustic medium (later also for elastic), starting with one-way wavefield reciprocity theorems. They showed how to reconstruct the reflection response in the case of deterministic as well as white-noise sources in the subsurface. At about the same time, other authors derived similar relations using different techniques for applications in different fields (Weaver and Lobkis, 2001; Schuster, 2001; Derode et al., 2003; Snieder, 2004) and as a result different terminology was introduced for the same process. To avoid confusion, it was proposed, in a special issue of Geophysics dedicated to these techniques, to use the term Seismic Interferometry (SI) for reconstruction through crosscorrelation (Geophysics, July-August issue of 2006). One of the applications of SI for exploration is the reconstruction of the reflection response from the crosscorrelation of seismic background noise recorded at the surface. In the normal seismic processing scheme, such noise is discarded. Several authors have shown the successful application of SI for the reconstruction of surface waves from seismic noise (Campillo and Paul, 2003; Sabra et al., 2005; Shapiro et al., 2005). In the following sections, we present results from the application of SI to seismic background-noise data recorded in a desert area. The aim of the experiment was to reconstruct reflection records that can be used in exploration, for example in frontier exploration areas. FIELD EXPERIMENT DESCRIPTION In 2005, Shell carried out a small field experiment to test the applicability of SI with seismic background noise for reconstruction of the reflection response. The experimental set-up consisted of 17 standard industry 3-component geophones arranged in a single line. The geophone spacing was 50 m and the time-sampling rate was 4 ms. The array was planted in a desert area. The 1 particular site was chosen so that there would be an active seismic survey available along the line to allow for verification of the reconstructed results and that the cultural noise was minimal during the recording of the background noise. Standard exploration equipment was used, which allowed for a maximum record length of 70 s. The background noise recording was then interrupted for 30 s to store the already acquired record. To be able to reconstruct the reflection response of a medium from crosscorrelation of noise, one needs time series at least of the order of hours. For this reason, 524 70-seconds records were acquired amounting to about 10 hours of seismic background-noise data. RECONSTRUCTION OF THE REFLECTION RESPONSE To reconstruct the reflection response from the recorded background-noise data, we used the SI relation as derived in Wapenaar and Fokkema (2006) using two-way wavefield reciprocity. In the frequency domain this relation reads 2R {