Earthquake forecasting through a smoothing Kernel and the rate-and-state friction law: Application to the Taiwan region

We applied two forecasting models for spatio-temporal distribution of seismicity density rate based on the smoothing Kernel function (SKF) and rate-and-state friction law (RFL) in Taiwan region to test their feasibility. Earthquake catalog from 1973 to 2007 was used to build up a time-independent forecasting model through the SKF. Coulomb stress changes imparted by the M≥4.5 earthquakes from 2008 to 2009 were calculated in order to propose a time-dependent model by the RFL. The distribution of M≥3.0 earthquakes from 2008 to 2009 is forecasted to examine our results. For SKF and RFL models, the percentage of forecasted earthquakes located within the 50% of the study area with high calculated seismicity rate are 82% and 72%, respectively. Results show that both of the models represent good abilities in earthquake forecasting. We further propose another model by combination of these two approaches. The rate could further reach to 84 %. It will be useful for seismicity forecasting in the near future.