It isn't ruled out. The LIGO team have carefully looked at, modelled and empirically measured all sorts of possible noise effects that could give the observed signature in both detectors within a few ms of each other. This includes a battery of other instruments which monitor all sorts of things, including seismic activity, that are used to veto sections of data so that they aren't used in any analysis.
The confidence with which they are suggesting they believe this event was not caused by any noise source that they have considered or measured, is quantified by saying that noise would only produce a signal that looked like the merger of two black holes in both detectors once every 200,000 years of LIGO observation.
Note that this "false alarm probability" does not rule out some other cause besides the merging of black holes for the event. As far as I know, nobody has suggested a credible alternative that can mimic the chirp signature of merging black holes.
However, if anybody has any ideas, then now is the time to publish and make predictions for the distribution of such events and their strengths, since I imagine it is quite likely that LIGO will publish results from several months of observing shortly.
As far as your specific suggestion goes, would a deep mantle earthquake produce almost simultaneous signatures in detectors thousands of km apart that lasted for a fraction of a second and swept across an order of magnitude range in frequencies? Given p-wave velocities of 5 km/s or so, there would have to be an incredible lack of any wave dispersion/refraction/diffraction for a wavepacket to arrive in such a short space of time after travelling thousands of km?