Time-Lapse Full-Waveform-Inversion under Variable Source And Receiver Coupling - a Deconvolution Method

Abstract

Summary Monitoring based on geophysical tomography is an important tool for characterising changes in the subsurface induced by natural or anthropogenic processes. Such processes typically cause only small changes in the data, such that standard monitoring techniques are highly susceptible to changes introduced by variable acquisition conditions. If these changes are known, they can be removed from the data prior to inversion. However, if these changes are unknown, additional parameters have to be introduced to the inversion problem to account for (variable) coupling conditions. In this study we investigate the effects of variable coupling conditions on standard time-lapse seismic Full-Waveform inversion procedures. We show that the double-difference method is unsuited to be applied in scenarios, in which acquisition conditions change in between monitoring surveys. We address this shortcoming and propose a novel monitoring approach based on (de-)convolution. This approach is then applied to a seismic data set acquired at the Svelvik CO2 field laboratory and compared with the established sequential and double-difference approaches. Compared to these approaches, our novel method shows superior waveform fit and fewer artificial structures in the time-lapse tomograms.

Authors 2

1. H. Söding Aachen

   RWTH Aachen University · ETH Zurich

   Affiliation as printed

   ETH Zürich

   RWTH Aachen University

2. H. Maurer

   ETH Zurich

   Affiliation as printed

   ETH Zürich