High-resolution FWI: Changing the Way we Image and Interpret Seismic

Abstract

Abstract The emergence of Full Wavefield Inversion (FWI) methods represents a paradigmshift with the potential to revolutionize the way we image and interpretseismic. Advances in algorithms and computing technology are now making FWIpractical, making it possible to construct detailed subsurface property modelsthat adequately explain the full bandwidth of the seismic data. We recentlydeveloped new concepts and algorithms that allow us to run 3D FWI using muchhigher frequencies - up to 45 Hz for one of the examples we will be discussing- than what has been shown in most published FWI studies to-date (usually lessthan 8–10 Hz). High-resolution FWI products can be viewed as 3D volumes of welllogs, and their availability creates new opportunities in the way we interpretand use seismic data. We expect FWI to impact applications at all businessstages, ranging from improved structural imaging during exploration to detailedreservoir description and characterization for development andproduction. Introduction Seismic imaging has been advancing rapidly over the last two decades, transitioning from post-stack time migration to pre-stack depth migration andfrom ray-theoretic algorithms (Kirchoff migration) to one-way wave-equationsolvers (wave-equation migration), and finally to more accurate reverse-timemigration (RTM) methods. This progression has been motivated by the businessclimate, with exploration moving to more and more challenging environments(e.g. Gulf of Mexico sub-salt) and the demand for more and more accuratereservoir characterization to maximize and maintain production from discoveredfields. The most recent, and perhaps ultimate, step in this evolving sequenceis Full Wavefield Inversion (FWI), which has rapidly become a focal area forresearch and development within the geophysical community and industry. Onecould argue that the FWI approach represents the dream of every geoscientist:generate detailed subsurface property models of the earth that can accuratelyreproduce, through simulation, the seismic data used to construct them. Unlikeevolution in the migration algorithms, which has been continuous andincremental, FWI represents a change in philosophy, both in the way we imagethe seismic data, and also in how we interpret the resulting products. Although much more research and development will be necessary to capitalize onthe full potential of FWI, we feel that several of the main difficulties(primarily overcoming the computational requirements necessary to implement andapply the method) have been overcome. We are already using FWI to generatesubsurface models with the resolution and quality to impact business decisions. This presentation will provide an overview of the method, some of the advancesthat are making it feasible, and real data examples to illustrate itspotential.