Review of time-reverse applications to locate and characterize seismicity in geothermal reservoirs

In the last two decades, enhanced computational resources and deployments of dense station networks inspired new kinds of methods that exploit the full amplitude and phase information of the recorded waveform. These waveform-based or data-driven methods do not rely on the identification of the phases in the seismograms and thus typically cope better with lower signal-to-noise ratios than ray-based methods. They are thus especially well-suited for applications in geothermal reservoirs where large numbers of microseismic events may occur quasi-simultaneously.
Time-Reverse Imaging (TRI) is one waveform-based method for locating and characterising seismic events. The recorded seismic traces are time-reversed and re-inserted at their respective recording positions for a numerical simulator solving the elasto-dynamic wave equation. The time-reversed wavefield is back propagated using a suitable velocity model until it converges towards the true source locations. Applying suitable imaging conditions, i.e. observing the energy density during the back propagation, results in high-accuracy source images.
In recent years, passive seismic datasets have rapidly increased in volume and resolution. This demands the development of automatic tools to fully analyse these datasets. TRI has the potential to completely, efficiently and accurately locate and characterise most seismic events in these large datasets. The increased data volume and larger computing capacities resulted in the generation of high-resolution three-dimensional velocity models and subsequently it became feasible to apply TRI as a routine tool in many applications.
Here we report recent advances in investigating the potential of TRI to locate and characterise seismic events are reported. In particular, the influence of the station network configuration, the velocity model and the signal-to-noise ratio on the source-location accuracy is reviewed. The method's ability to locate and characterise seismic events in the context of geothermal reservoirs, which often experiences a large number of small magnitude seismic events, is shown. Recent advances such as the determination of the time-dependent moment tensor using TRI are also reported. The applicability of the method is demonstrated through examples from real case applications.