Calibrating and testing of the forecasts of seismic hazard for planned mining sequences

The forecast of seismic hazard in mines depends on the planned mining sequence and therefore it is required to: 1. Model the changes in stresses and strains associated with future mining. 2. Transform these changes to the parameters of expected potentially damaging seismic events (location, time, size, mechanism). This modelling of expected seismicity has to be calibrated. It needs to be shown that the recorded seismic response to the mining in the past can be replicated using a numerical model. The seismic hazard calculated for future mining steps also needs to be tested against the observed seismicity after the planned mining is completed. The same mathematical framework can be used for both the calibration and the testing of seismic hazard forecasts. The area skill score (Zechar & Jordan 2008) is adopted to assess the match between the location of significant seismic events and calculated hazard maps for the past mining steps (calibration) and forecast maps for the future mining steps (testing). The 3D rotation angle (Kagan 2007) is used to compare the source mechanisms of recorded significant seismic events with the expected mechanisms for the past and future mining steps. The seismic events affecting the poor performance of the forecast both in terms of location and source mechanisms can guide possible adjustment to the input parameters of the model (e.g. orientation of in situ stress, failure criteria) and help to improve the forecasts. The suggested approach of calibrating and testing of seismic hazard forecasts is illustrated using data from Renison mine, Australia.