Combining Envelope Inversion and Full Waveform Inversion for Velocity Model Building Using OBN Data

Abstract

Oil and gas constitute a significant part of marine resources, whose exploration and production routinely rely on geophysical seismic data to improve existing databases. Traditional seismic datasets have suffered from extensive ambiguity due to the combined effects of thermal fluids, fractures, shallow velocity anomalies, and mid-deep layer bottom structure. Actual situation displays that despite several previous reprocessing efforts, the quality of seismic data in some ambiguous areas has only been partially improved, while the resolution issues remain unresolved. Given the poor seismic imaging quality and extraordinarily low signal-to-noise ratio data, it is challenging to determine the accurate structural features. To tackle this challenge, we combine the analysis of wavefield attenuation characteristics and high-precision velocity modeling as well as inversion methods to depict the subsurface areas. We aim at provide a validate high-quality seismic data processing and inversion method to determine an accurate velocity model to support the interpretation of actual geological data. We introduce the low-frequency reconstructed envelope inversion and investigate GPU/CPU paralleled full waveform inversion (FWI) method. We demonstrate the efficacy of our method by using both synthetic data and field data from a certain sea area in China. Numerical results indicate that a combination of low-frequency reconstructed envelope inversion and full waveform inversion can effectively improve the accuracy of the existing velocity data for more than 25%.