Stability and Uncertainty of Full Moment Tensor Solutions for M < 3.5 Induced Earthquakes

The increase in earthquakes associated with industrial activities has created a need to investigate and characterize the source physics of induced seismicity. Many techniques and approaches are available to determine representative source parameters of these events. For M > 3.5 events, high quality seismic data from regional networks can be used to provide reasonable estimates of moment tensor solutions. In this investigation we explore various techniques and datasets to constrain full moment tensor solutions of M < 3.5 induced events, expanding upon the approach developed by Guilhem et al., 2014. Small magnitude events recorded by local seismic networks can yield good quality data with distinct body wave and converted phases depending upon the velocity structure and frequency range. Generating synthetic seismograms or Green's functions to accurately model these high frequency phases can be challenging. To investigate the variability associated with the choice of Green's functions, we test available codes to see how well they capture body wave phases. Other stability and uncertainty measures include the F-test, Jackknife test, residual bootstrap, and Network Sensitivity Solution, (Ford et al., 2009; Ford et al., 2010). Additional datasets to constrain the full moment tensor solution include P-wave first motions and amplitude ratios.