Quantifying subsurface geophysical properties changes using double-difference seismic-waveform inversion with a modified total-variation regularization scheme

Abstract

Quantifying time-lapse changes of subsurface geophysical properties is crucial for many applications, such as monitoring for oil/gas production, for geologic carbon storage, and for enhanced geothermal systems, etc. We develop a new double-difference acoustic-waveform inversion method and a new double-difference elastic-waveform inversion method using a modified total-variation regularization scheme for accurate estimation of subsurface geophysical properties changes. The method jointly inverts time-lapse seismic data for changes of geophysical properties in target monitoring regions. Our new waveform inversion algorithms incorporate a modified total-variation regularization scheme consisting of two regularization terms: an L2 norm term and an L1 norm total-variation term. We employ an alternating minimization method to decouple our new waveform inversion with the modified total-variation regularization into two minimization subproblems to improve the robustness of waveform inversion. We use seismic-waveform inversion with a modified total-variation regularization scheme to produce an accurate baseline geophysical model using the baseline seismic data and apply our new double-difference seismic-waveform inversion to time-lapse seismic data to quantify time-lapse changes of geophysical properties. Our new double-difference waveform inversion algorithm not only preserves sharp interfaces of the target monitoring regions but also reduces inversion artefacts outside the target monitoring regions. We use synthetic time-lapse seismic data to validate the improvement of our new methods. Our numerical results show that our new double-difference acoustic- and elastic-waveform inversion methods significantly improve the accuracy of time-lapse seismic data inversion compared to other inversion methods.