A sensitivity analysis of nullspace shuttling to illumination, anomaly magnitude, and size using sparse acquisition geometries

Kimberly Pike, Scott Keating, Kristopher A. Innanen

Time-lapse targeted nullspace shuttling is a post-FWI optimization method which explores alternative parameter models within the inversion nullspace while preserving the FWI data misfit. When applied to time-lapse problems, time-lapse nullspace shuttling seeks the minimum model difference consistent with the data, making it a promising tool for "sparse monitoring scenarios". In this study, we assess the sensitivity of time-lapse nullspace shuttling to three key factors: (1) the magnitude of true velocity changes, (2)the lateral extent of the induced anomaly, and (3) illumination based on sparse acquisition geometry. Results show that time-lapse nullspace shuttling recovers large-magnitude time-lapse anomalies even under sparse monitor geometries, however it struggles to isolate small-extent, low-magnitude anomalies. We introduce illumination-based and stratigraphic masking strategies to restrict the spatial domain of the nullspace shuttling objective function.