Data-Driven, Direct Rock-Physics Inversion of Pre-Stack Seismic Data

Miguel Angel Corrales Guerrero, Matteo Ravasi, Hussein Hoteit

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The inversion of petrophysical parameters from seismic data represents a fundamental step in the reservoir characterization framework. However, the non-linearity of the rock-physics models that relate petrophysical properties to seismic pre-stack amplitudes makes such an inversion challenging. We propose a hybrid approach, where data-driven basis functions are learned from well-logs to directly link band-limited petrophysical reflectivities to pre-stack seismic data. Petrophysical parameters are subsequently obtained by means of regularized post-stack seismic inversion. By performing two modeling steps at training time and a single inversion step at inference time, our method aims to be more efficient and robust than conventional two-step inversion workflows. The proposed method is tested on a synthetic dataset from the Smeaheia reservoir model. Numerical results show that porosity is the best-inverted rock property, followed by water saturation and clay content. Moreover, the method is shown to be also applicable in the context of reservoir monitoring to invert time-lapse, pre-stack seismic data for water saturation changes.
Original languageEnglish (US)
Title of host publication83rd EAGE Annual Conference & Exhibition
PublisherEuropean Association of Geoscientists & Engineers
DOIs
StatePublished - 2022

Fingerprint

Dive into the research topics of 'Data-Driven, Direct Rock-Physics Inversion of Pre-Stack Seismic Data'. Together they form a unique fingerprint.

Cite this