Induced Seismicity Monitoring Tools for the Geothermal Environment

Abstract

The monitoring of induced seismicity has become an important objective as geothermal and other energy production operations have moved closer to population areas. Fluids withdrawn from or injected into the subsurface cause changes in the local stress field, which can result in the generation of earthquakes, either on new or pre-existing faults. Examples of such earthquake activity include The Geysers geothermal field, where induced seismicity has occurred for decades, the 2008-2009 earthquake swarms near the Dallas/Ft. Worth airport, and the 2010-2011 earthquake swarms in central Arkansas, possibly including a magnitude 4.7 event. The latter two examples are likely attributed to fault reactivation from subsurface water injection. The majority of seismic monitoring tools were developed for regional and teleseismic distances. Without adaptation, these tools are not well suited for the small-scale heterogeneity and the higher frequency content data observed at the reservoir scale. In order to monitor for and possibly prevent unwanted induced seismicity, it is necessary to develop new or adapt and refine existing tools. An initial step in monitoring for induced seismicity is to develop a seismic velocity model. We have adapted a regional scale joint inversion technique that includes important reservoir scale features such as travel-time prediction methods that are not limited to layered structures or surface receivers and 3-D nonlinear velocity tomography with geostatistical constraints. To aid development of a starting model for the inversion, we investigated the use of a passive seismic technique known as seismic noise tomography, a method traditionally employed at regional distances to calibrate the velocity model between two stations. We developed an event detection algorithm specifically designed to identify small seismic signals in noisy data. Event locations are then determined by a specially modified version of the Grid-Search Multiple Event Location (GMEL) software originally developed to locate nuclear testing around the world. In this paper, we present the development of this software and examples of its application on induced seismicity data sets.