Premium
Accurate tremor locations from coherent S and P waves
Author(s) -
Armbruster John G.,
Kim WonYoung,
Rubin Allan M.
Publication year - 2014
Publication title -
journal of geophysical research: solid earth
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.983
H-Index - 232
eISSN - 2169-9356
pISSN - 2169-9313
DOI - 10.1002/2014jb011133
Subject(s) - seismology , geology , episodic tremor and slip , slip (aerodynamics) , subduction , coherence (philosophical gambling strategy) , geodesy , geodetic datum , physics , tectonics , quantum mechanics , thermodynamics
Abstract Nonvolcanic tremor is an important component of the slow slip processes which load faults from below, but accurately locating tremor has proven difficult because tremor rarely contains clear P or S wave arrivals. Here we report the observation of coherence in the shear and compressional waves of tremor at widely separated stations which allows us to detect and accurately locate tremor events. An event detector using data from two stations sees the onset of tremor activity in the Cascadia tremor episodes of February 2003, July 2004, and September 2005 and confirms the previously reported south to north migration of the tremor. Event detectors using data from three and four stations give S and P arrival times of high accuracy. The hypocenters of the tremor events fall at depths of ∼30 to ∼40 km and define a narrow plane dipping at a shallow angle to the northeast, consistent with the subducting plate interface. The S wave polarizations and P wave first motions define a source mechanism in agreement with the northeast convergence seen in geodetic observations of slow slip. Tens of thousands of locations determined by constraining the events to the plate interface show tremor sources highly clustered in space with a strongly similar pattern of sources in the three episodes examined. The deeper sources generate tremor in minor episodes as well. The extent to which the narrow bands of tremor sources overlap between the three major episodes suggests relative epicentral location errors as small as 1–2 km.