Simultaneous waveform inversion for velocity, density and source moments with application to seismic-while-drilling

Jinji Li, Scott Keating, Roman Shor, Kristopher A. Innanen

Full waveform inversion (FWI) is an optimization-based approach to estimating the subsurface parameter model that minimizes the difference between synthetic and real data iteratively. In practice, incomplete acquisition and illumination of the subsurface are strong limiting factors. Adding data corresponding to new and independent ray paths as input could lead to significant increases in the reliability of FWI models. In principle, seismic-while-drilling (SWD) can supply these additional ray paths but introduce a new suite of unknowns, namely precise source locations and radiation characteristics. Here we formulate a new elastic FWI algorithm in which source positions and radiation patterns join the velocity/density values of the grid cells as unknowns to be determined. We then carry out a synthetic feasibility study in which such incompletely-known sources are included along a plausible well-trajectory through a simulated model, around which seismic receivers are placed in various configurations. This SWD-FWI is optimized with a Truncated Gauss-Newton algorithm. The subsurface model and source properties (P-wave velocity, density, and three independent 2D moment tensor values) are recovered and analyzed. The analysis suggests that the participation of SWD improves the accuracy of FWI models, especially in density inversion. The inversion of elastic properties shows a similar improvement when different drilling paths participate, which is different than expected. Such results indicate further related study is required to provide more comprehensive radiation patterns of the SWD sources.