Contributions to the Risk Assessment of Natural and Induced Seismicity: Producing High-Resolution Catalogs and Improving Risk-Based Decision-Making

Earthquakes can pose a serious threat to people and their environment. Natural earthquakes have the potential to be disastrous and mostly occur unexpectedly. Induced earthquakes caused by anthropogenic activities are commonly less disastrous, but can be a great nuisance and are increasingly attracting public attention. The physical processes behind seismicity are not understood well enough to predict natural earthquakes or to completely control induced earthquakes. Estimates about the future occurrence of earthquakes are typically very uncertain, which makes it difficult to use these estimates for mitigation strategies. These restrictions partially arise from the limited resolution of available data and analyses. This thesis will address these limits two-fold. First, it presents a holistic approach for natural earthquake sequences that translates the typically very low earthquake occurrence probabilities into risk estimates—these might reach intolerable levels and justify mitigation actions; in the broadest sense, these may save lives. Secondly, it presents a fundamental approach to produce a high-resolution earthquake catalog of consistent quality using a sensitive detection method and dedicated post-processing; this approach detects very small earthquakes and offers new possibilities for seismicity analyses. The first contribution in this thesis describes an alternative risk mitigation strategy based on probabilistic risk forecasting and cost–benefit analysis, which can support objective short-term evacuation decisions. We apply this approach to a scenario earthquake sequence that simulated a repeat of the 1356 Basel earthquake (∼Mw6.7), one of the most damaging events in central Europe. Based on this simulated sequence, we explore how to possibly provide decision support and information throughout an earthquake crisis. We forecast earthquake consequences, specifically quantified in terms of human loss. The final cost–benefit analysis adds value beyond making probabilistic forecasts: it provides objective statements that may justify evacuations. Compared to an earlier study, we incorporate recent developments in short-term earthquake forecasting and more detailed settlement data to permit spatial forecasts and district-wise decision-making. These new ingredients permit an increased spatial resolution for the decision support and have the potential to improve the cost-effectiveness of mitigation actions. For instance, the occurrence of an M5.5 earthquake or larger would justify evacuations in the central part of the city of Basel, Switzerland. To deliver such supportive information in a simple form, we propose a warning approach in terms of alarm levels. The second contribution in this thesis describes the development and analysis of a consistent high-resolution earthquake catalog of induced seismicity in the Enhanced Geothermal System (EGS) in Basel. The Basel EGS was a deep geothermal energy project. Seismic monitoring at the Basel EGS site has been running for more than a decade. Yet, the details of the long-term behavior of its induced seismicity remained unexplored because a seismic event catalog that is consistent in resolution and in magnitudes did not exist. Such details are essential for developing guidelines and mitigation strategies on how to safely operate and terminate injection activities. We reinvestigate the induced sequence in detail and search for the smallest earthquakes in the available seismic recordings of the injection and long-term period. We detect these “buried” earthquakes with a sensitive matched filter technique based on cross-correlation. Due to quality deficits in the recordings, we apply advanced techniques, including a machine learning method, to obtain consistency in terms of a high detection sensitivity and robust magnitude estimates. The new catalog contains more than 280 000 earthquakes down to Mw−1.5, which have never been considered before. These events offer a much higher temporal resolution of seismicity and gain new insights into the seismic behavior of the EGS. In the injection period, we resolve temporal variations of seismicity parameters that indicate a hazardous modification of the seismogenic behavior in the EGS. In the long-term period, we find a preferential temporal clustering of earthquakes. We further demonstrate the new capabilities offered by the high-resolution catalog. We find a breakdown in the Gutenberg–Richter scaling during the reservoir stimulation, which we show is partially caused by a method-independent detection limitation during high event rates, and likely has additional physical reasons. This scaling break has implications for the design of advanced mitigation strategies and must be carefully considered in future EGS projects. Further, we already provide insights into the long-term behavior of the induced Basel sequence by investigating the temporal clustering and whether surface waves of remote earthquakes modulate seismicity by dynamic triggering. Our analyses reveal characteristic foreshock activity increases in the earthquake clusters, but no indication for remote dynamic triggering. In summary, this thesis contributes novel approaches to seismological research, pro- vides useful input for further studies, and may stimulate discussions among researchers, decisions makers, and the public. The individual contributions of this thesis improve specific aspects of seismic risk assessment and could eventually facilitate the reduction of seismic risk. The first contribution gives an idea of the decision support that can be currently provided by research and may specifically raise awareness of the omnipresent seismic risk in Basel; it could lay the groundwork to inform the public about their own current seismic risk level. The second contribution will give the opportunity to study the sequence in unprecedented detail and resolution, which might help to better understand the physical processes in a deep geothermal reservoir. These insights may improve the way in which deep geothermal projects are brought to success, for example, by better evaluating reservoir creation to guarantee safety for society. The presented method to produce consistent high-resolution catalogs may facilitate better-calibrated seismicity forecasting models and to simultaneously forecast more timely. Viewed optimistically, both contributions of this thesis combined facilitate a new compelling pursuit: a more accurate and timely short-term risk assessment to further improve decision support