Automatic processing of induced seismicity at the geothermal reservoirs Landau and Insheim
Besides political and striking uncertainty, induced seismicity is a fundamental risk to the success of geothermal projects. The geothermal reservoirs Landau and Insheim are located on the western rim of the Upper Rhine Graben where the highest fluid temperatures in Germany provide a good foundation for geothermal energy production while the complex tectonic structure provides a difficult set-up for geophysical investigations. In this thesis I advance automatic processing algorithms and apply them to assess characteristics of induced seismicity at the two geothermal reservoirs. Key objectives are to complete existing detection catalogs, determine the reliability of automatic processing results and to identify characteristics of induced earthquakes at the geothermal reservoirs which in future may be linked to operational parameters of the power plants. Challenging the completeness of existing catalogs a complementary offline similarity detection approach which considers multiple master events per reservoir is proposed. The second algorithm is an autoregressive prediction (AR)-Akaike Information Criterion (AIC)-cost-function arrival time determination approach which is straightforward trainable for different datasets on different scales and able to reliably pick robust seismic P- and S-phase arrival times on single- and multi-component data. The resulting hypocenter offsets standard deviations are approximately 1 km horizontally as well as vertically. Finally from hypocenter relative relocations, I am able to identify a spatial separation of the Insheim earthquake clusters retrieved by the similarity detection algorithm. Two orientations of seismicity are found to coincide with the stress regimes, firstly a south-east to north-west orientation during graben activity, and secondly a rather south-to-north orientation of the recently more dominant stress regime from Alpine denudation.