The ML 3.5 earthquake sequence induced by the hydrothermal energy project in St. Gallen, Switzerland

Starting in March 2013, the geothermal project of the city of Sankt Gallen, Switzerland, has drilled through 4 km of sedimentary rocks in the Swiss Molasse Basinin order to find and exploit hydrothermal aquifers in the Mesozoic sediments. In a large-scale 3D seismic survey, the project operators identified a nearly 30 km long N-S striking segmented fault zone in the Mesozoic sediments. Based on the apparent lack of recent seismic activity, they concluded that the fault zone was not active and drilled into this target of potentially enhanced permeability. In July 2013 a testing and stimulation program began in the Malm sediments. A small-scale fresh water injectionon July 14 was followed by two acid stimulations. A low level of seismicity that strongly correlated with the testing program was observed by the Swiss Seismological Service (SED) on a dedicated network of 10 surface stations and one shallow borehole station. The seismicity during this period did not exceed magnitude ML1.2 and was judged to be well within the expected range. When operators were preparing for an airlift test, methane gas was released into the borehole from an unknown source around noon on July 19. The pressure at the wellhead rose rapidly, and operators decided to pump water and heavy mud down the well. Even though wellhead pressure decreased steadily, seismicity started to increase suddenly at 7 pm (UTC) on July 19. Although the traffic light system designed by the operators was triggered in the early phase of the seismicity increase, operators found themselves forced to continue well control instead of stopping the pumps. During this period, the seismicity intensified and culminated in a ML 3.5 event at 3:30 (UTC) on July 20 that was widely felt in the area. Yet, the SED received only a small number of reports on minor non-structural damage. In the following hours, the operators were able to stabilize the well and flare the methane in a controlled manner. Seismicity decreased rapidly within a few days but two weeks later was still far from reaching the background level. Here we report on the results of our analysis of the induced seismic sequence at Sankt Gallen: ML 3.5 event initiated near the borehole, had a comparatively low stress drop (3.5 bar) and a rupture length of ~1.1 km. Peak ground motions observed for the ML 3.5 eventare very similar to the ones observed in the ML 3.4 event induced in 2006 in Basel. Yet, macroseismic intensities in St. Gallen only reach IV (EMS) versus V (EMS) in Basel. Precise earthquake relative locations indicate that seismicity extends bi-laterally from the injection point, following the trend of the mapped fault segments. Fault plane solutions of the two largest events indicate a left lateral strike slip fault whose orientation agrees well with the aftershock locations and the imaged fault zone. The St. Gallen sequence shows the highest seismic productivity per injected fluid volume when compared to other injection-induced sequences, and challenges proposed relations between injected fluid volume and maximum observed magnitude or between hydraulic energy and released seismic energy.