Regional Characteristics of Observable Foreshocks

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 com-pleteness level. Foreshocks are identified using three approaches: a magnitude -depen-dent 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 earth-quakes in the magnitude range M*-2 <= M< M*, in which M* is the mainshock magni-tude. 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 rel-atively high probabilities of having at least one foreshock in Italy ( similar to 43%-56%), com-pared 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 proba-bilities 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.

Automated summary

Measures of foreshock occurrence in eight different regions were examined using earthquake catalogs. Three different approaches were used to identify foreshocks: a fixed-time windowing method, a nearest-neighbor clustering method, and a variable-time windowing method. The method with fixed-time windows consistently yielded higher counts of foreshocks. There were differences in foreshock probabilities between regions and the method used, but overall, regions with high background seismicity had low probabilities of detectable foreshocks.