Earthquake forecast/prediction techniques applied to real-time testing experiments

We have applied an earthquake clustering epidemic model to real-time data of the Istituto Nazionale di Geofisica e Vulcanologia (INGV) at the Italian Earthquake Data Center for short-term forecasting of moderate and large earthquakes in Italy. The model uses earthquake data only, with no explicit use of tectonic, geologic, or geodetic information. The whole procedure is being tested in real time at the CSEP testing Center operating at ETH in Zurich. The same model has also been applied at the New Zealand Seismological Observatory-Wellington (Geonet) for forecasting in the New Zealand region. Forecast verification procedures applied to this stochastic triggering epidemic model have been carried out in forward-retrospective way both on the 2006-2007 INGV data set and January 2006-April 2008 Geonet data set, making use of statistical tools as the log-likelihood ratio, the Relative Operating Characteristics (ROC) diagrams, the Molchan error diagrams, the probability gain and the R-score diagrams. These procedures show that the clustering epidemic model performs up to several hundred times better than a simple random forecasting hypothesis. The seismic hazard modeling approach so developed, after a suitable period of testing and refinement, is expected to provide a useful contribution to real time earthquake hazard assessment, even with a possible practical application for decision making and public information.