Abstract

The 14 November 2001 Kokoxili, Tibet, earthquake (Mw 7.8) ruptured ∼400 km of the western Kunlun fault in northern Tibet. We apply two inversion methods to P and SH waves recorded by the Global Seismographic Network to recover the spatial and temporal history of the rupture process. The observed surface rupture consists of two strike-slip segments, offset by an extensional step-over. Little surface rupture was observed in the graben system in this step-over, which is approximately 45 km long and over 10 km wide. Our results imply that the rupture did not jump this large gap, but that the rupture was continuous through the graben. The earthquake began with a small strike-slip subevent, presumably along the westernmost of the two segments. This was followed 5 sec later by a subevent of about the same magnitude but having an oblique mechanism with a large normal-faulting component. The likely location for this subevent is within the graben. This oblique-slip event probably enabled transfer of the rupture onto the main Kunlun fault. The main moment release began ∼18 sec after rupture initiation and propagated over 350 km eastward along the Kunlun fault. Slip on this segment was very heterogeneous, averaging ∼2 m for the first 200 km, followed by a sharp increase to a maximum of 7.5 m within the next 50 km, and then a rapid decline. The average rupture velocity along the main segment was high (∼3.6 km sec-1) and probably exceeded the local shear-wave velocity. The Mw 7.8 Kokoxili earthquake had a longer surface rupture and faster average rupture velocity, radiated more energy, and probably had a lower average fracture energy than the November 2002, Mw 7.9 Denali fault (Alaska) earthquake. Evidence suggests that the velocity of the rupture front dropped significantly after passing the point of maximum slip, implying a large difference in frictional strength between the two ends of the fault.

Online material: Waveform fits to teleseismic body waves.

You do not currently have access to this article.