The hypocenter (location underground where the earthquake started) is fairly deep (~26 km according to the USGS) so hopefully the rupture didn't reach the surfa

The hypocenter (location underground where the earthquake started) is fairly deep (~26 km according to the USGS) so hopefully the rupture didn't reach the surface (it's ambiguous but maybe 30% likely for an earthquake of this size with a hypocenter at that depth). If it didn't, then the likelihood of a damaging tsunami is far less because there isn't a big displacement at the seafloor. Interestingly, from the location it looks like it's right at the northern edge of the rupture patch of the 2010 Maule earthquake (8.8), which was at the northern end of the 1960 ~9.4 earthquake (biggest on record). These earthquake rupture patches are visible here[1]. The idea is that as the oceanic plate subducts, the upper part of the subduction zone/fault/megathrust is locked, i.e. the upper and lower plates are stuck together above 20-30 km, and elastic stress and strain accumulate slowly. Then, this stress and strain is sporadically released as earthquakes, but these don't rupture the entirety of the fault. Instead, they happen in patches but an earthquake on one part of the fault adds more shear stress to the adjacent parts of the fault that didn't rupture. This basically steps the clock forward on the adjacent part, so it ruptures next. In this way, the displacement accumulated along a big fault over centuries or millennia is released in a relatively shorter timespan, then a long period of quiescence sets in. That's a theory, anyway, and it's almost certainly right in some places, some of the time. A big research question for some (not necessarily funding agencies) is figuring out whether the shear stress released in these big earthquakes is a big or a small part of the accumulated background stress (stress is nigh impossible to measure directly). If most of the shear stress is released, it'll be a long time before another earthquake can occur on a given fault segment, but if only a little bit is released, then the potential energy is still there for another earthquake and it can happen again quicker than the full plate reloading time. From a statistical perspective, a full stress release may correspond to periodic earthquakes while a partial release may correspond to a more random (i.e. Poisson) recurrence pattern. If anyone is interested in this stuff, let me know here or email me (in my profile) and I can try to write up a blog post with it, and maybe with some Monte Carlo simulations to illustrate. [1]: ftp://ftp.ingv.it/pub/alessio.piatanesi/PAPERS/2011_NGEO_Chile.pdf