Full Waveform Inversion

Overview

Full waveform inversion (FWI) is a seismic imaging technique developed by Patrick Lailly and Albert Tarantola in the early 1980s. It estimates the earth's structure by matching predicted waveforms (synthetic seismograms) to observed waveforms "wiggle for wiggle." This method can accurately represent recorded seismic waveforms if the physics of wave propagation is correctly applied to the true Earth model. While conceptually simple, practical application of FWI had to wait for the evolution of computing resources over a couple of decades. Now, with the widespread availability of high-performance computing, FWI is used in all areas of seismology, from short-period surveys to long-period tectonic studies.

FWI relies on synthetic waveform modeling, which is enhanced by massively parallel forward solvers operating in the time or frequency domain on 2D or 3D data. As hardware continues to improve, it is now possible to model and estimate the complete anisotropic and attenuating subsurface.

The slideshow includes the workflow for FWI data processing.