This dissertation presents the first integrated geophysical interpretation that jointly analyzes time-lapse seismic and marine controlled-source electromagnetic (CSEM) data at the Sleipner CO2 storage site, the world’s first and longest-operating co...
This dissertation presents the first integrated geophysical interpretation that jointly analyzes time-lapse seismic and marine controlled-source electromagnetic (CSEM) data at the Sleipner CO2 storage site, the world’s first and longest-operating commercial-scale offshore CO2 storage project. The workflow combines seismic and CSEM inversion with an anisotropy assumption to more reliably interpret subsurface properties, including the evolution of the CO2 plume and anisotropy at the Sleipner site.
First, anisotropic time-lapse full-waveform inversion (FWI) based on the global correlation norm (GCN) is applied to both 2D and 3D seismic vintages, and the results are compared with those obtained from isotropic time-lapse FWI. Since 2D FWI affords greater computational efficiency than 3D FWI, a multi-scale frequency strategy using frequencies up to 60 Hz is employed, allowing an assessment of the benefits of incorporating higher-frequency components. In contrast, the 3D FWI is performed at up to 30 Hz, and the comparison confirms that 4D FWI monitoring results provide more accurate and reliable large-scale velocity updates than 2D time-lapse results. As a result, all anisotropic inversion results exhibit distinct P-wave velocity changes due to CO2 injection, compared to the isotropic case, thanks to the consideration of the anisotropy parameter ε. Moreover, synthetic data generated from the anisotropic results show improved alignment with the observed data, especially for reflections from the caprock and storage layer.
Next, an initial resistivity model is constructed for the CSEM inversion. A previous study employed a well-log-based background model; however, upper stratigraphic information is missing at Sleipner due to partial loss of available well logs. To address this limitation, the P-wave velocity model obtained from the 3D anisotropic FWI in 2008 is used, the same year as the 2D CSEM survey, and is converted to resistivity using rock-physics relationships. The Raymer–Hunt–Gardner (RHG) algorithm is first applied to transform the P-wave velocity into porosity, and Archie’s law is subsequently used to convert porosity into electrical resistivity. All physical parameters and constants required for these transformations are calibrated using well-log data from the Sleipner site.
Subsequently, the constructed initial resistivity model is used to perform a 2.5D anisotropic CSEM inversion on a single 2D marine CSEM dataset acquired in 2008. In this context, the 2.5D algorithm simulates 3D EM source behavior within a 2D subsurface geological model, thereby improving computational efficiency. For the inversion process, the “NGI25EM” software developed at the Norwegian Geotechnical Institute (NGI) is utilized, updating the vertical (ρv) and horizontal (ρh) resistivities, as well as the electrical anisotropy ratio (ρv/ρh). The inversion results show that the shale- and mudstone-dominated caprock exhibits higher electrical resistivity and stronger electrical anisotropy than the saline-filled sand formation beneath it. Compared with a well-log-based initial model, the FWI-based initial model improves structural continuity and reduces shallow artifacts in the inversion results. The boundary depths between the caprock and the storage formation are also better constrained, as verified through well-log quality control (QC).
Finally, through the integrated interpretation of the anisotropic FWI and CSEM inversion results with the well-log dataset from the injection and observation wells, a consistent geophysical understanding of the Sleipner site is achieved. The caprock interval identified from seismic velocity and gamma-ray logs corresponds to a high-resistivity, strongly anisotropic zone in the CSEM results. Both the seismic anisotropy (ε) and electrical anisotropy (ρv/ρh) exhibit higher values due to the shale-rich caprock's inherent properties. They gradually decrease downward into the storage formation, confirming that anisotropy is a dominant characteristic of the overburden sequence above the storage layer. Thus, this highlights the importance of accounting for anisotropy when interpreting subsurface properties at the Sleipner site.
Consequently, the integrated approach presented in this study provides more accurate inversion and monitoring results, addressing the limitations of previous Sleipner studies. This thesis establishes its originality by (i) being the first to perform both seismic and CSEM inversions directly on the field dataset and jointly interpret them at the Sleipner site, (ii) analyzing the seismic and electrical anisotropies of the sedimentary layers, and (iii) reconstructing P-wave velocity and resistivity models with more improved accuracy and geophysical consistency, compared with the existing studies. Furthermore, the proposed workflow provides a transferable framework for future research on deep-learning-based joint inversion and multi-physics monitoring at other commercial-scale sites, such as the Smeaheia project.