[1] Analysis of broadband seismograms from the 15 July 2003 large earthquake (M 7.6) in the central Indian Ocean reveals an unusual source process. The source duration of longer than a minute is more than twice as long as expected from earthquake scaling relations, yet $80% of the moment release occurred in two energetic asperities near the end of the rupture. These two asperities were located in lithosphere with an age of 7 Ma or greater. A previous study has suggested that strike-slip earthquakes in oceanic lithosphere having much longer than expected source durations also have a slow, dissipative rupture process characterized by low radiated seismic energy (and therefore low apparent stress). We find no evidence for a slow rupture process to the 2003 earthquake. Instead, the long duration appears to be due only to nucleation close to the actively spreading Carlsberg Ridge, in lithosphere younger than 7 Ma. Younger oceanic lithosphere may be able to generate small to moderate earthquakes but be unable to sustain slip in a large event due to steady release of strain in aseismic creep events. Large strike-slip earthquakes within oceanic lithosphere may occur only in the central portions of long transform faults or in intraplate regions, rupturing energetic asperities like those that failed in the mid-Indian Ocean earthquake and leading to the observation that oceanic strike-slip earthquakes have the largest apparent stresses among the global population of shallow earthquakes.

[1] Earthquakes associated with the creation of new oceanic crust are difficult to locate precisely because they normally occur far away from seismic stations. We assume that epicenters of these events can be constrained by seafloor topography and use the Harvard centroid moment tensor (CMT) focal mechanism of an interplate earthquake to relate the event to either a transform fault (a parallel in a coordinate system defined by the pole of rotation between two plates) or a mid-oceanic ridge (a meridian). Using arrival times from the International Seismological Center (ISC) Bulletin, we have relocated about 1500 globally distributed events for the time period 1976-1998. The ISC locations can be as much as 70 km away from appropriate topographic features. Our new locations support that large interplate earthquakes occur on or very close to the primary bathymetric features. For earthquakes with unknown mechanism or magnitude below the CMT threshold or predate the CMT catalog, we use a simultaneous Joint Hypocenter Determination procedure with respect to the already relocated Master Events (JHDME). We invert for earthquakes first for the Romanche Fracture Zone (RFZ) and then for a region extending from 30°N to 30°S. Seismicity agrees well with the main active ridge or transform. However, a small number of events might be related to the secondary bathymetry features. The improved locations for mid-oceanic earthquakes based on bathymetry could help improve our understanding of the oceanic crustal and deep Earth structure, seismic slip and budget constraint along plate boundaries, and dynamics of plate tectonics.

[1] Mid-oceanic interplate earthquakes are difficult to locate accurately because they normally occur far away from land-based seismic stations. Use of water-borne T waves recorded by autonomous underwater hydrophone (AUH) arrays records an order of magnitude more highly accurate regional low seismicity along the north Mid-Atlantic Ridge than the International Seismic Centre (ISC). Even though the physical meaning of an AUH locations is still not well known, AUH's small location errors are important for better constraining mid-oceanic earthquakes. Comparison of such AUH locations with those in ISC and Harvard centroid moment tensor (CMT) location catalog, and relocated ones based on the high-resolution bathymetry and teleseismic P phases, is made in this study. AUH locations are used as a reference to compare the teleseismically determined locations. For large earthquakes with known focal mechanisms, we find that relocated locations agree with AUH ones better than with ISC. We also note that the centroid vectors from relocated epicenters are usually larger than AUH centroid vectors. The relocated epicenters and AUH locations lie in similar azimuthal directions to the associated CMT epicenters. The larger relocated and AUH centroid vectors (than the error ellipses of AUH, CMT, and relocated ones combined) might be explained by the fault rupture process. For smaller events, relocated location confidence ellipses are usually large enough to cover AUH locations and their error ellipses. Overall, the highly accurate AUH locations can be used to confirm the mid-oceanic earthquake hypocenters and seismicity characteristics and for detail studies of the low-level seismicity associated with the plate motions.Citation: Pan, J., and A. M. Dziewonski (2005), Comparison of mid-oceanic earthquake epicentral differences of travel time, centroid locations, and those determined by autonomous underwater hydrophone arrays,