Seismic technologies

[SE-01] Robust time-lapse full waveform inversion using boundary integral representation: numerical examples for borehole and ambient seismic noise monitoring

*Shohei Minato¹ (1. Delft University of Technology (Netherlands))

Full waveform inversion (FWI) has a potential to obtain high-resolution images of the subsurface. When FWI is applied to time-lapse data with permanent receivers, accuracy in source positioning and knowledge outside the target area affect the detectability of small temporal changes. In this study, we develop a new approach for robust time-lapse FWI using a boundary integral representation of wavefield. The representation uses an array of receivers as a reference and extrapolates the observed waves at the reference array. This enables performing FWI without accurate information of source geometry, source wavelet, or structure outside a target area. We consider monitoring experiments using two vertical boreholes and surface sources, and numerically show that the new approach estimates accurately time-lapse changes at a target area when there is an uncertainty in an initial velocity model at a non-target shallow layer. We also investigate effects of source positioning errors and those of non-repeatable source wavelets. Finally, we discuss benefit of the proposed approach in surface wave monitoring using ambient noise. The approach will be crucial in enabling an accurate and stable estimation of temporal changes in the subsurface structures.