Abstract

Quantitative seismic interpretation can be used to identify lithology and detect petroleum accumulations by integrating rock properties and attributes derived from advanced seismic inversion methods with existing petrophysical data and geological knowledge. We use quantitative seismic interpretations for detection of shallow biogenic gas accumulations in the Qaidam Basin, China, employing an integrated workflow that incorporates petrophysical data, seismic attribute analysis, Constrained Simultaneous Inversion (C-SI) and Bayesian-based Support Vector Machine (B-SVM) inference. Previous petrophysical studies have shown that it is challenging to effectively identify gas-bearing intervals using parameters such as impedance, Poisson’s ratio and porosity, because the reservoir sediments are unconsolidated and at shallow depths. The resistivity well-log response remains as an effective tool for estimating gas saturation and identifying gas-bearing intervals. In this study, we propose the use of the petroleum pore-volume, which is defined as the product of reservoir porosity and gas saturation, to detect biogenic gas accumulations seismically. Rock properties inferred from seismic inversion, such as compressional velocity (Vp), shear velocity (Vs) and density, cannot be used directly for petroleum pore-volume estimation. Therefore, we employ a Bayesian-based support vector machine approach to cross-link well-log properties, seismic AVO attributes and seismic rock properties to quantitatively predict petroleum pore-volume in 2D and 3D seismic dataset. Because seismic information is crucial to statistical inference, we propose C-SI to infer the Vp, Vs and density from seismic elastic impedance gathers, which can be generated from seismic gathers using a traditional recursive seismic inversion method. The C-SI procedures use the Interior-Point algorithm to optimize and solve elastic impedance equations. The Interior-Point method is a popular method for handling constrained non-convex, non-linear optimization problems that involve simultaneously inverting the seismic properties with thousands of seismic samples. This case study indicates that the integrated study workflow is useful for quantitatively predicting petroleum pore-volume, especially in the depth-domain, and that it is an excellent potential indicator for biogenic gas accumulations in complicated geological settings.