Recognizing societal influences in earthquake geohazard risk perception with explainable AI while mitigating risks through improved seismic interpretation

Abstract

As public awareness of climate-change-related weather events and temperature anomalies increases in the United States, citizens’ attitudes toward climate change should be studied, including the effects of climate mitigation strategies and their associated risks such as induced seismicity. Using social survey data and earthquake records, an explainable artificial intelligence method (SHAP) is employed to investigate factors that are important in explaining individual perceptions and assessments of future induced seismicity. Focusing on Oklahoma, where the majority (greater than 75%) of earthquakes are induced, SHAP reveals that personally experiencing earthquakes is a significant factor in the respondents’ past and future perception of earthquake frequency. Realizing that induced seismicity plays a strong role in individual perceptions of earthquake frequency, it is increasingly important to mitigate this geohazard, which is expected to increase with climate mitigation strategies, whether they are carbon capture and sequestration or geothermal in nature. To this end, seismic attribute methods to improve subsurface characterization, particularly for fluid migration pathway identification, are examined using data from a carbon sequestration project in northwest Montana in the Kevin Dome area. While broadband and multispectral coherence do not improve the identification of faults and fractures, in this case, aberrancy (the third derivative of structure) successfully highlights lineations because it is more susceptible to detecting flexures in seismic data. Based on this result, we strongly encourage interpreters to include aberrancy in their reservoir analysis to identify and mitigate fluid migration that may result in induced seismicity.