Velocity model building and imaging in the presence of shallow gas
A.R. Ghazali, R.J.J. Hardy, T. Konuk, R.I. Masiman, K. Xin, M.H. Mad Zahir and M.F. Abd Rahim
Journal name: First Break
Issue: Vol 34, No 10, October 2016 pp. 79 - 84
Info: Article, PDF ( 801.05Kb )
Price: € 30
Multiple shallow layers of low saturation gas charged sands cause poor seismic imaging and hamper exploitation of many oil and gas discoveries in SE Asia. These gas ‘clouds’ are characterized by bright high amplitude reflectors of low P-wave velocity beneath which the seismic reflections have delayed P-wave travel times, reduced bandwidth, reduced reflected and transmitted amplitudes and associated phase distortion (Figure 1). All gas clouds are not equal, the appearance of gas clouds varies worldwide and within regions. A single gas-charged sand may sometimes have almost no effect on P-wave propagation; a shallow region of limited spatial extent may sometimes be undershot resulting in an interpretable image or multiple shallow layers can result in complete ‘wipeout’ of the final image. We believe that both anelastic losses and multiple scattering phenomena (indicated by rays on Figure 1) contribute to weak reflected energy beneath the gas-charged sands. A similar problem in the North Sea led to the development of multi-component seabed seismic solutions utilizing converted waves to penetrate beneath the gas (Berg et al., 1994). These solutions have been successfully applied to the largest of the Malaysian producing fields and are the PETRONAS preferred technical solution (e.g. Radzi et al., 2015). As with many problems, a mixture of acquisition and processing related solutions produces the optimal result but sometimes processing alone can produce a cost-effective result. Despite cost reductions and technical advances, converted wave imaging is still expensive, difficult and time consuming. In this paper we attempt to build an efficient and accurate workflow for velocity model building of P-wave data in order to best exploit the latest imaging algorithms. We use a mixture of synthetic and real data examples from offshore Malaysia to illustrate this difficult and unresolved problem.