Enhancement of the short-term probability of large earthquakes with a foreshock model and verification test

This study attempts to estimate the short-term probabilities of large earthquakes in an offshore area of northeast Japan, where long-term probabilities are assessed based on earthquake intervals. Previous studies report foreshock activity in the area, which could be used for the estimation. From this point of view, a hazard function was constructed based on the concept of "potential foreshocks" of the study area. A total of 14 earthquakes that occurred between 1976 and 2000 are employed in an assessment of models. The probability of a target earthquake occurring at a point in the time-space domain depends on the number of small earthquakes in the vicinity of the point. The parameters for defining potential foreshocks are magnitude, spatial extent, and lead-time to the point, which is optimized by a maximum probability procedure. The most effective hazard function is achieved based on foreshocks of magnitude 4.5 and larger within 1 day and 20 km. A maximum probability gain of more than 20,000 is obtained if there are two or more foreshocks. We started verification tests applying the model to data from January 1, 2001 to date, but conclusive results were not yet obtained due to the small sample size of target events.