Improvements in seismic data quality can significantly enhance hydrocarbon production, motivating the investigation of methods to acquire more accurate and reliable data. In many cases there will be considerable uncertainty in reservoir properties and in the level of error in the data. We present an approach for determining the potential value of competing seismic survey methods to improve knowledge of reservoir properties to inform decisions related to reservoir management. Monte Carlo simulations using an earth model based on a Gulf of Mexico site, and seismic error models, provide statistical estimates of the ability of seismic amplitudes to infer porosity and reservoir thickness. Bayesian decision analysis methods then facilitate the optimization of an infill drilling program and allow the quantification of the economic value of the different seismic data sets.
Seismic data play an important role in reservoir characterization, in application to tasks ranging from the selection of a drilling target to the direct detection of fluids in reservoir formations. Detailed and successful reservoir characterization requires accurate and reliable seismic data for optimal results. Improvements in such factors as signalto- noise ratio, bandwidth, streamer cable position, or resolution can help improve seismic image quality and the results of interpretation methods such as amplitude variation with offset (AVO) in ways that significantly improve knowledge of reservoir structure or the variation of porosity and thickness, for example. For these reasons, improving data acquisition to increase seismic accuracy can be very beneficial. The influence of acquisition geometry on seismic imaging, for example, can be quantified to allow a quantitative comparison of the benefits of alternative acquisition geometries. In particular, model-based methods for these comparisons allow survey design to be customized to specific sites. However, these approaches tend to rely on comparisons of a small number of models, perhaps only one, to compare the influence of changes in acquisition procedures on image quality, examining factors such as spatial resolution. This neglects uncertainty in the model or acquisition technology. Bayesian decision theory is a useful tool for addressing this problem. Given statistical models for the relationships between the quantities of interest (e.g., porosity) and some observable form of data (e.g., seismic amplitude), application of Bayes’ Theorem provides a means for determining the probability of accurately making decisions based on such data. Recent work has applied an extension of this, Bayesian value of information (VOI) theory, to investigate the role of seismic data in the process of selecting a drilling location (Stibolt and Lehman 1993; Waggoner 2002; Ballin et al. 2005). However, many practical problems include the selection of multiple drilling targets, and much of the previous work has relied on expert assessment rather than quantitative, model-based calculation of the accuracy of seismic data. Houck (2004) presented model-based calculations to quantify the improvement in seismic images resulting from a hypothetical improved streamer positioning system determined the economic value of improved data. Bickel et al (2006) quantify the reliability and value of seismic information in the context of a 3D land example.
