Selecting ground-motion models developed for induced seismicity in geothermal areas

We present a case study of the ranking and weighting of ground-motion prediction equations (GMPEs) for seismic hazard assessment of enhanced geothermal systems (EGSs). The study region is Cooper Basin (Australia), where a hot-fractured-rock project was established in 2002. We test the applicability of 36 GMPEs based on stochastic simulations previously proposed for use at EGSs. Each GMPE has a set of corresponding model parameters describing stress drop, regional and local (near-surface) attenuation. To select suitable GMPEs for Cooper Basin from the full set, we applied two methods. In the first, seismograms recorded on the local monitoring network were spectrally analysed to determine characteristic stress and attenuation parameters. In a second approach, residual analysis using the log-likelihood (LLH) method was used to directly compare recorded and predicted short-period response spectral accelerations. The resulting ranking was consistent with the models selected based on spectral analysis, with the advantage that a transparent weighting approach was available using the LLH method. Region-specific estimates of variability were computed, with significantly lower values observed compared to previous studies of small earthquakes. This was consistent with the limited range of stress drops and attenuation observed from the spectral analysis.