High-Resolution Long-Term and Short-Term Earthquake Forecasts for California

摘要: We present two models for estimating the probabilities of future earth- quakes in California, to be tested Collaboratory Study Earthquake Predictability (CSEP). The first is a time-independent model adaptively smoothed seismicity that we modified from Helmstetter et al. (2007). provides five- year forecasts earthquakes with magnitudes M ≥ 4:95. show large tend occur near locations small 2 events, so high- resolution estimate spatial distribution obtained numerous events. further assume universal Gutenberg- Richter magnitude distribution. In retrospective tests, Poisson distri- bution does not fit observed rate variability, contrast assumptions current earthquake predictability experiments. therefore issued using better- fitting negative binomial number second time-dependent epidemic-type aftershock sequence (ETAS) (2006) and next-day 3:95. this model, forecasted sum background (propor- tional rate) expected triggered events due all prior earthquakes. Each triggers increases exponentially its decays time according Omori-Utsu law. An isotropic kernel density aftershocks (M ≤ 5:5) while larger quakes, smooth early forecast later parameter values by optimizing find short-term realizes probability gain about 6.0 per over model. Online Material: Identification explosions ETAS parameters.