Investigation into the use of 2D elastic waveform inversion from look‐ahead walk‐away VSP surveys

Abstract

A B S T R A C TPore pressure in sediments beneath salt in the Gulf of Mexico varies widely creating a potentially dangerous and difficult drilling challenge. Estimating elastic parameters of sediments beneath salt is key to the prediction of pore pressure and reducing the drilling risk in exiting the base of the salt. In this paper we investigate the ability of 2D waveform inversion to recover the elastic parameters in the sedimentary layer beneath the salt from a walk-away VSP (vertical seismic profile) carried out with the receivers in the salt, with the objective of estimating pore pressure at the base of the salt (to be estimated using traditional methods). We propose an effective method for performing the inversion and apply this method to a blind test of a large and realistic synthetic dataset. To facilitate the design of a VSP survey suitable to this type of inversion we also present an analysis into the effects of the receiver array receiver spacing. It is shown that the resulting velocity estimates are sufficiently accurate to predict the pore pressure within the limits required for drilling.

Time‐lapse full waveform inversion of vertical seismic profile data: Workflow and application to the CO2CRC Otway project

“…Roberts et al . [] chose the velocity parameterization for FWI of VSP data due to the fact that transmitted arrivals are dominant in terms of energy. Following the same logic, we choose to parameterize the inversion in seismic velocities (compressional V P and shear V S ) and density ρ .…”

“…Analysis of the scattering patterns by different inhomogeneities [Tarantola, 1986] suggests that the impedance-density parameterization is more suitable for small reflection angles, while velocity-density parameterization is more appropriate for large reflection angles and transmitted geometries. Roberts et al [2008] chose the velocity parameterization for FWI of VSP data due to the fact that transmitted arrivals are dominant in terms of energy. Following the same logic, we choose to parameterize the inversion in seismic velocities (compressional V P and shear V S ) and density ρ.…”

“…Owusu et al [2016] performed FWI of VSP data with a wavelet derived from far-field measurements of an airgun array. Pratt [1999] and Roberts et al [2008] introduced different source wavelet inversion algorithms in frequency domain and successfully used them for the inversion of experimental data. In our current study, we estimate the wavelet directly from the vertical component of the offset VSP data by stacking a number of traces of the downgoing wavefield.…”

Wavenumber Sampling Issues in 2.5D Frequency Domain Seismic Modelling

“…It was shown by WILLIAMSON and PRATT (1995), in a critical review of acoustic modelling, that the errors can become substantial in complex media and where multiple, overlapping arrivals are present. Despite such limitations the pseudo 2.5D method and the 2D method remain popular (mainly because of the simplicity of 2D modelling) and are used in many inversion algorithms (CHOI et al 2008;ROBERTS et al 2008;SEARS et al 2008). Inversion of ground penetrating radar data is also based on 2D modelling and simple corrections to adjust the recorded (2.5D or 3D) data to 2D wavefield characteristics (ERNST et al 2007).…”

Introduction: Towards a full waveform inversion

“…Roberts, Singh and Hornby (2008) apply a 2D elastic waveform inversion to invert looked‐ahead walk‐away VSP surveys. The goal is to obtain a better estimation of the elastic parameters beneath the salt and then improve the pore pressure prediction to reduce the drilling risk.…”