BIO & ABSTRACT
Jim Fowler has been a QI geophysicist for 25+ years, beginning with Jason Geosystems in 1998. Since then, his work has focused on seismic inversion, AVO studies, and all aspects of reservoir geophysics. Most recently, he spent 14 years in multiple European countries with Wintershall Dea, culminating as Geophysics Coordinator for Norwegian Exploration, thereby gaining currency in wider geophysical applications (acquisition, processing and interpretation). Currently, Jim is based in Houston as Sr. Project Development Geophysicist for Qeye.
Finding Signal in the Noise: Using multiples and mode conversions to improve AVO inversion
Klaus Rasmussen, Evan Mutual, Henrik Hansen, Raul Cova, Bill Goodway, Rob Ross, Jim Fowler*, Qeye Labs
(*presenter)
ABSTRACT: Seismic reservoir characterization in geologic settings with coals, anhydrites, and gas clouds is known to be challenging due to transmission effects and non-primary energy introduced by strong velocity contrasts. The problem of these high-reflectivity events masking the amplitude response of underlying reservoirs is well known and understood to be particularly troublesome for AVO inversion workflows. In this presentation, we compare the standard conventional convolutional AVO inversion approach to a new workflow that uses a 1D tau-p domain wave equation as the forward modelling operator in an effort not only to more accurately model the Full Waveform input seismic data, but to use the short period multiples and mode conversions to improve AVO inversion results.
DISCUSSION: Complex geologic settings with high impedance contrasts are known to introduce errors in conventional seismic images due to transmission effects and non-primary energy, often considered unwanted noise. However, removing these “noise” elements of the wavefield (i.e., short period multiples or mode conversions) from the observed raw seismic data is a very complex task and risks damaging primary energy. Applying reservoir characterization workflows to this seismic data, particularly seismic inversion, is very challenging as the presence of interbed multiples and mode conversions contributes to a highly complex seismic image that cannot be accurately modeled with conventional seismic convolutional theory. Overlapping wavefield contributions at both near and far offsets will often degrade the intercept and gradient (and overall AVO) solutions.
In this presentation, we compare a conventional AVO inversion workflow, which relies on the standard convolutional model, with a new 1D tau-p wave equation method which more accurately models the input seismic data by including short period multiple and mode conversion waveforms in the solution. This equation assumes that the subsurface is slowly varying horizontally, but accurately models multiples, mode conversions, and can incorporate velocity dispersion.
A case study is shown where the amplitude responses of hydrocarbon reservoirs are complicated by the proximity of other high-impedance contrast lithologies (clays, anhydrites, etc.). The case study highlights differences between the conventional AVO approach and the new tau-p full-wavefield AVO (FW-AVO) inversion. The new approach shows significant improvement in forward modelling results, with similar uplift achieved in the inversion of observed seismic. The derived elastic properties (acoustic impedance and Vp/Vs) from the FW-AVO inversion show a better match with well logs than the conventional convolutional AVO inversion.
CONCLUSION: This paper introduces a method for AVO inversion that inverts more of the recorded wavefield than the standard convolutional model, by including short-period (pegleg) multiples and mode conversions. The tau-p FW-AVO method addresses conventional imaging limitations, particularly in environments with coals, anhydrite stringers and other high reflectivity boundaries. In the case study shown, we demonstrate first with forward modelling and then with inversion how accounting for these wavefield phenomena in the seismic model can dramatically improve the understanding of observed seismic and the accuracy of inverted elastic properties, thereby reducing uncertainties in reservoir characterization.
