Full waveform inversion in the time lapse mode applied to CO₂ storage at Sleipner

Abstract

Carbon capture and storage is a viable greenhouse gas mitigation technology and the Sleipner CO2 sequestration site in the North Sea is an excellent example. Storage of CO2 at the Sleipner site requires monitoring over large areas, which can successfully be accomplished with time lapse seismic imaging. One of the main goals of CO2 storage monitoring is to be able to estimate the volume of the stored CO2 in the reservoir. This requires a parametrization of the subsurface as exact as possible. Here we use elastic 2D time‐domain full waveform inversion in a time lapse manner to obtain a P‐wave velocity constrain directly in the depth domain for a base line survey in 1994 and two post‐injection surveys in 1999 and 2006. By relating velocity change to free CO2 saturation, using a rock physics model, we find that at the considered location the aquifer may have been fully saturated in some places in 1999 and 2006.

Quantitative seismic characterization of CO2 at the Sleipner storage site, North Sea

“…Boait et al (2012) and Bergmann and Chadwick (2015), using volumetric estimations, derive similar conclusions about the fluid distribution; i.e., the saturation is most likely uniform in high saturation zones and following patchy mixing laws elsewhere. Golding and Huppert (2011) agree with Queißer and Singh (2013) that saturation is probably greater than 30%. Ghosh et al (2015) quantify the CO 2 saturation with a maximum of 20% for the uniform saturation case and with a maximum of 80% for the patchy mixing case, with an uncertainty of approximately 5%-7%.…”

“…On the 2006 vintage, Williams and Chadwick (2012) also estimate very high saturation values. Queißer and Singh (2013) are using FWI results to estimate CO 2 saturation via Gassmann fluid substitution. On the 2006 vintage, they get saturations ranging from 20% (deeper layers) to 90% (870 m deep layer), but P-wave velocities are very low, approximately 1200 m∕s.…”

“…By exploiting all information contained in the seismic signal, the method has the highest potential in terms of resolution to locate and provide quantitative estimates of velocity changes caused by the presence of injected CO 2 . At Sleipner, attempts to derive high-resolution quantitative maps of velocity properties have been made (Queißer and Singh, 2013;Romdhane and Querendez, 2014;Raknes et al, 2015). Although the method can locate and image the CO 2 plume, seen as property changes induced by the CO 2 injection, the quantification of the uncertainty in the reconstructed parameters is challenging.…”

Increasing CO<sub>2</sub> flux at Pisciarelli, Campi Flegrei, Italy

“…In line with this prospect is a clear seismic velocity decrease since 2012 (Zaccarelli and Bianco, 2017), which could be due to, for instance, a softening bulk or increase in CO 2 saturation in the CF aquifer (Queißer and Singh, 2012) that may also explain the strong seismic attenuation observed (De Siena et al, 2017b).…”

Interferometric full-waveform inversion of time-lapse data

“…This makes FWI an ideal choice for inverting for timelapse changes in the subsurface medium. Recently, many applications of 4D FWI to recover temporal model variations have appeared in the literature (Routh et al, 2012;Queisser and Singh, 2013;Raknes and Arntsen, 2014) even though it is sensitive to non-repeatability errors (Asnaashari et al, 2012). Maharramov et al (2016) simultaneously inverted the baseline and monitor datasets, while also imposing a model-regularization constraint to counter non-repeatability problems.…”