ExxonMobil and Input/Output jointly recorded a densely sampled 3C/3D wave-spread near Houston, Texas to characterize ground roll for an area that was assumed to have a simple near surface. The resulting tests demonstrated that, at this location, ground roll is significantly more complicated than a simple Rayleigh wave. Detailed analysis of the data is continuing.
Ground roll presents a serious data quality problem for many land seismic surveys as it generally exhibits both long wavelength and high amplitude. With conventional seismic acquisition techniques, ground roll is poorly attenuated by geophone arrays and is often under sampled leaving it severely aliased in both shot and receiver domains. These properties can make ground roll mitigation and recovery of the underlying reflection data very difficult. The most basic mitigation techniques of bandpass filter and ''surgical mute'' also unfortunately remove reflection energy. FK filters and phase-shift techniques (Kim, 1998) have limited effectiveness if the ground roll is aliased, dispersive, scattered, or composed of several velocities. Even cascaded filters with different velocities can be ineffective. There has also been a revival of interest in mitigation based on polarization (Diallo, 2005), especially with the recent development of 3C MEMS sensors (Burch, 2004; Ronen, 2005; and Kendall, 2005). This paper will demonstrate some of the characteristics that make ground roll mitigation difficult and suggests that adequate sampling may be essential for effective processing operations.
In July 2005, ExxonMobil Upstream Research Company and Input/Output jointly acquired a highly sampled 3C/3D wave spread at the ExxonMobil research facility south of Houston, Texas. From previous studies, it was known that ground roll was very slow ,so large offsets would not be required for adequate characterization. The test was designed around a densely sampled 3C/3D wave spread and 2D line (Figure 1) using 3C Vectorseis sensors on the surface. The 2D inline spacing was 5m and the wave spread was laid out with inline and cross-line spacing of 5m. In addition, one grid line was sampled at a 1m inline interval. All inline sensors were carefully aligned to be parallel with the 2D line. Good coupling was assured by forcing each Vectorseis sensor into a smaller diameter pre-drilled hole. To improve the characterization, down hole data were simultaneously acquired. Four holes were instrumented with 3C geophones at a 1.5m interval from 1.5m - 30m. There was also a pre-existing instrumented borehole with 3C geophones from 15m - 305m at a 15m interval. At one of the 30m boreholes, 3C geophones were also planted in open holes at 30cm intervals from 0 - 1.2m. In order to provide additional information regarding the near surface, 13 CPT (Cone Penetrometer Technology) tests were acquired to determine the lithology and mechanical properties of the near surface. Seismic Cone Penetrometer Technology shear check shots were acquired to help quantify the near surface shear velocity. Nineteen different tests were acquired over a four day All sensors were live for all tests and sampled at 1ms with recording filters at Out and 3/4 Nyquist.
