Earthquake fault superhighways D.P. Robinson ⁎, S. Das, M.P. Searle Department of Earth Sciences, Parks Road, Oxford, OX1 3PR, UK abstract article info Article history: Received 4 September 2009 Received in revised form 9 December 2009 Accepted 26 January 2010 Available online 1 February 2010 Keywords: Seismology Super-fast rupture Earthquake hazard Rupture velocity Motivated by the observation that the rare earthquakes which propagated for significant distances at supershear speeds occurred on very long straight segments of faults, we examine every known major active strike-slip fault system on land worldwide and identify those with long (N 100 km) straight portions capable not only of sustained supershear rupture speeds but having the potential to reach compressional wave speeds over significant distances, and call them “fault superhighways”. The criteria used for identifying these are discussed. These superhighways include portions of the 1000 km long Red River fault in China and Vietnam passing through Hanoi, the 1050 km long San Andreas fault in California passing close to Los Angeles, Santa Barbara and San Francisco, the 1100 km long Chaman fault system in Pakistan north of Karachi, the 700 km long Sagaing fault connecting the first and second cities of Burma, Rangoon and Mandalay, the 1600 km Great Sumatra fault, and the 1000 km Dead Sea fault. Of the 11 faults so classified, nine are in Asia and two in North America, with seven located near areas of very dense populations. Based on the current population distribution within 50 km of each fault superhighway, we find that more than 60 million people today have increased seismic hazards due to them. © 2010 Published by Elsevier B.V. 1. Introduction It is now understood (Das, in preparation, review paper this issue) that one of the parameters of the earthquake source that controls the resulting damage is the earthquake rupture propagation speed. Calculations showed that the ground shaking during an earthquake is directly related to the velocity and acceleration of the rupture front (Madariaga, 1983). Theoretical models of earthquake rupture, starting in the early 1970s, showed seismologists that earthquake ruptures could not only exceed the shear (S) wave speed but could even reach the compressional (P) wave speed (e.g. Kostrov and Das, 1988). For a long time the only example of an earthquake rupture exceeding the shear wave speed was the 1979 Imperial Valley earthquake (Arch- uleta, 1984). As the distance during which supershear rupture occurred was short (b 10 km) and the data was not unambiguous, many scientists did not fully accept that supershear rupture was possible in nature. But since 1999, examples of supershear rupture speeds on the north Anatolian fault in Turkey (Bouchon et al., 2001), and near-compressional wave speed ruptures on the Kunlun fault in Tibet (Robinson et al., 2006; Vallée et al., 2008; Walker and Shearer, 2009) and on the Denali fault in Alaska (Walker and Shearer, 2009) convinced seismologists that such speeds were truly possible. At about the same time, laboratory experiments found rupture speeds, not only exceeding the shear wave speed (Xia et al., 2004) but reaching the compressional wave speed (Xia et al., 2005), and provided additional impetus for this acceptance. Fracture mechanics shows that it is only strike-slip faults that are likely to reach such fast speeds, and no observation of supershear rupture during a dip-slip earthquake, such as those that occur in subduction zones, for example, has ever been found. The fact that the 2001 Kunlun, Tibet earthquake, the 2001 Denali earthquake and the 1999 Izmit, Turkey earthquake propagated at high rupture speeds for significant distances (about 100 km for the Kunlun fault) on the remarkably straight segments of the earthquake fault, can be explained from theoretical considerations. Earthquakes start from rest and need to propagate for some distance to reach their maximum speed (e.g. Kostrov and Das, 1988). Once the maximum speed is reached, the earthquake could continue at this speed, provided the fault is straight, and no other barriers exist on it. Faults with many large changes in strike, or large step-overs, would thus be less likely to reach very high rupture speeds as this would cause rupture on such faults to repeatedly slow down, before speeding up again. The ground velocities and accelerations generated by such super-fast ruptures can be five times or much larger than subshear ones at a distance of 30 km from the fault, due to the generation of S and Rayleigh wave Mach cones (shock-type waves, analogous to the “sonic boom” from supersonic aircraft), whose amplitudes decay much more slowly with distance than usual spherical waves (Bernard and Baumont, 2005; Dunham and Bhat, 2008). The 2001 Kunlun earthquake is particularly interesting, as in addition to a variety of different seismic studies (Robinson et al., 2006; Vallée et al., 2008; Walker and Shearer, 2009) that have concluded Tectonophysics 493 (2010) 236–243 ⁎ Corresponding author. E-mail address: davidr@earth.ox.ac.uk (D.P. Robinson). 0040-1951/$ – see front matter © 2010 Published by Elsevier B.V. doi:10.1016/j.tecto.2010.01.010 Contents lists available at ScienceDirect Tectonophysics journal homepage: www.elsevier.com/locate/tecto