Measures of foreshock occurrence are systematically examined using earthquake catalogs for eight regions (Italy, southern California, northern California, Costa Rica, Onshore Japan, Alaska, Turkey, and Greece) after imposing a magnitude ≥3.0 completeness level. Foreshocks are identified using three approaches: a magnitude‐dependent space + fixed‐time windowing method, a nearest‐neighbor clustering method, and a modified magnitude‐dependent space + variable‐time windowing method. The method with fixed‐time windows systematically yields higher counts of foreshocks than the other two clustering methods. We find similar counts of foreshocks across the three methods when the magnitude aperture is equalized by including only earthquakes in the magnitude range M*−2≤ M< M*, in which M* is the mainshock magnitude. For most of the catalogs (excluding Italy and southern California), the measured b‐values of the foreshocks of all region‐specific mainshocks are lower by 0.1–0.2 than b‐values of respective aftershocks. Allowing for variable‐time windows results in relatively high probabilities of having at least one foreshock in Italy (∼43%–56%), compared to other regional catalogs. Foreshock probabilities decrease to 14%–41% for regions such as Turkey, Greece, and Costa Rica. Similar trends are found when requiring at least five foreshocks in a sequence to be considered. Estimates of foreshock probabilities for each mainshock are method dependent; however, consistent regional trends exist regardless of method, with regions such as Italy and southern California producing more observable foreshocks than Turkey and Greece. Some regions with relatively high background seismicity have comparatively low probabilities of detectable foreshock activity when using methods that account for variable background, possibly due to depletion of near‐failure fault conditions by background activity.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.