Towards Real-Time Double-Difference Hypocenter Relocation of Natural and Induced Seismicity

In order to assess the fault-geometry and the spatio-temporal evolution of natural and induced seismicity, high-precision (relative) micro-seismic hypocenter locations are key information. From such precise relative hypocenter locations, we can infer e.g. the spatial extent during a seismic sequence, the seismogenic volume affected by stimulation procedures as well as geometries (orientation, segmentation) of potentially activated faults. Additionally, in the case of induced seismicity, the spatio-temporal evolution of seismicity (e.g. migration velocities of seismicity, r-t-diagrams) can be indicative for fluid-flow processes and provides first-order estimates of hydraulic properties of the reservoir as well as on the existence of possible hydraulic connections. Information on spatial extent, geometries and the spatio-temporal evolution of seismogenic structures can help to improve the seismic hazard assessment of natural and induced seismicity in real-time or near-real-time.

However, to make prompt use of information provided by such high-precision hypocenter locations requires relative relocations computed in near-real-time. This can be rather challenging, especially at the beginning of a seismic sequence, when only little or no background seismicity is available for relative relocation. In addition, an automated relative relocation process requires differential times derived from precise and reliable (absolute) automatic picks as well as from waveform cross-correlation.

In this work, we present our strategy towards a near-real-time relative relocation procedure. The procedure follows the methodology described by Waldhauser 2009 (BSSA; doi:10.1785/0120080294) and combines differential times derived from automatic as well as manual picks with waveform cross-correlations measurements. Differential times of new events are inverted for relative locations with respect to a background reference catalog using the double-difference algorithm. We present results derived by a python-based prototype applied to natural and induced earthquake sequences. In addition, the prototype is fully implemented in a new SeisComP3 (SC3) module “scrtdd”, which allows the application in a full real-time environment, using detections and locations from various existing SC3 modules (“scautoloc”, “scanloc”, “screloc”) as input for relative relocation. We outline our implementation strategy, and compare SC3 results with results derived by our software for natural and induced earthquakes monitored be dense near-fault monitoring networks in the Valais (SW Switzerland), St. Gallen (Switzerland) and the Hengill Geothermal Field (SW Iceland).