Wavelet Transform and Hidden Markov Model Application for Unsupervised Clustering of Microseismic Events from CO2 Injection

ABSTRACT: Injection of CO2 increases pore pressures that trigger complex subsurface interactions, leading to slip or reactivation of pre-existing faults/fractures and/or creating fractures. Understanding the subsurface response is key to preserving the integrity of the storage site, optimizing the injection, and mitigating the risk of seismic hazards. Machine learning (ML) models have shown the potential to improve our understanding of complex earth systems by identifying subtle changes in patterns of seismic and microseismic data. Furthermore, ML has shown great promise in revealing patterns in time-dependent spectral properties of (micro-) seismic data to identify differences in subsurface response. In this study, a wavelet transform scheme is implemented to isolate high energy frequency bands within microseismic data from the Illinois Basin Decatur Project (IBDP). Subsequently, an unsupervised ML method is used to identify various subsurface responses attributed to CO2 injection at the IBDP. Specifically, the unsupervised method known as SpecUFEx is used to further reduce the dimensionality of the reconstructed high energy micro-seismic spectrograms, solve for characteristic fingerprints, and apply K-means clustering to group similar fingerprints. Finally, a comparison is made between our clustering results and an implementation of SpecUFEx where the full frequency spectrum of the microseismic data was used. 1 INTRODUCTION Carbon capture and storage (CCS) has emerged as a critical technology in the global effort to mitigate climate change by reducing carbon dioxide (CO2) emissions from industrial processes and power generation. The concept of CCS involves capturing CO2 emissions from large point sources such as power plants and industrial facilities, transporting the captured CO2 to suitable storage sites, and securely storing it underground in geological formations to prevent its release into the atmosphere. This approach not only helps reduce greenhouse gas emissions but also enables the continued use of fossil fuels while transitioning to cleaner energy sources. As nations strive to meet ambitious climate targets, CCS is increasingly recognized as a key component of the portfolio of solutions needed to achieve significant and sustained reductions in CO2 emissions on a global scale. However, one primary risk of CCS is induced seismicity. Injecting CO2 into subsurface formations increases pore pressure, leading to changes in local stress conditions and potentially triggering micro-seismic events within basement rock. Accurate monitoring and modeling are imperative for the success of CCS projects, ensuring efficiency, safety, and societal acceptance.