Characterizing meander belts and point bars in fluvial reservoirs by combining spectral decomposition and genetic inversion

Abstract

Accurately characterizing meander belts and point bars in fluvial reservoirs is significant for reducing drilling risks and improving the recovery of oil and gas. For this goal, we developed a new seismic inversion method that combines spectral decomposition and genetic inversion for improving vertical and horizontal resolution of seismic interpretation and readjusting tuning effects. This method takes advantage of multiple spectral-decomposition seismic cubes. Seismic traces, integrated traces (90°-phase data), sweetness, and root-mean-square volume attributes were input to consider information from a range of seismic data types. Tests using a sample conceptual model and a real-world dataset demonstrate that the proposed method could achieve satisfactory results, improve the resolution of inversions, and readjust tuning effects. The real dataset is a case study from the Chengdao Oilfield in the Bohai Bay basin in China and the Neogene Guantao formation is the interval of interest. Our study area covers approximately 85 km2, and the dataset used includes mainly 441 wells and a post-stack 3D seismic cube with a dominant frequency of 40 Hz. In the spectral-decomposition genetic inversion results, distributions of inversion values were in relatively good accord with the well logs, and several thin adjacent sandstone bodies that could not be distinguished in the conventional genetic inversion results could be distinguished distinctly. We also proposed a workflow for the characterization of fluvial meander belts and point bars by integrating inversion results and quantitative depositional models. According to the surface attributes and red-green-blue blending maps of multiple stratal slices calculated from the inversion cube, one main meander belt, eight narrow channel fills, and several overbank sandstone bodies were accurately imaged; most of these features were confirmed from well observations. Furthermore, 15 point bars and related abandoned channels were characterized in the main meander belt. The proposed inversion method can improve the resolution of seismic interpretations and the undertaken characterization of meander belts and point bars could provide an example for other fields with a similar dataset.