A quick look this week at the magnitude 6.5 earthquake that shook southern Mexico last Sunday. It caused a fair amount of shaking in Mexico City, and a few deaths, but apparently no major structural damage. The depth of the rupture – around 65 kilometres (40 miles) – means that the seismic energy released had spread out over a wider area by the time it reached the surface, reducing the maximum shaking close to the epicentre.

A tectonic map of the region shows that there is a subduction zone running along the southwest coast of Mexico, where the oceanic Cocos plate is being subducted eastward beneath the North American plate. The rupture depth of 65 km puts it below the crust of the overriding plate, so it must be associated with the subducting slab. The focal mechanism indicates that the earthquake was due to northeast-southwest extension (see my primer on interpreting focal mechanisms).

Map showing location of the M 6.5 earthquake that struck south of Mexico City on 11 December, its extensional focal mechanism and proximity to the subduction zone off the southwest Mexican coast.

This might seem a bit odd at first glance: subduction zones are all about accommodating plate convergence, so a compressional, thrust focal mechanism would seem more likely. However, while that would certainly be the case for an earthquake that occurred on the subduction megathrust itself, this is not the only place in a relatively cold, subducting plate where earthquakes can occur if stress is being applied to the interior somehow. One mechanism that causes extensional earthquakes within the downgoing slab is if it is being bent; for example, you often see earthquakes with normal focal mechanisms on the outer rise of a subducting plate, where the plate is just starting to bend as it is pushed beneath the overriding plate into the mantle. This earthquake was 180 kilometres behind the trench, so was obviously not associated with this initial bending. However, gravity, GPS and seismic studies of this subduction zone have found that there is probably further bending of the slab beneath Mexico. The width of the locked zone associated with the subduction thrust, where elastic strain is being built up that will mostly be released in future megaquakes, is about twice as wide (200 km or so) as it usually is, suggesting that the Cocos plate is being subducted at a very shallow angle beneath southwest Mexico, before it eventually steepens again about 275 km away from the trench. As the image below, taken from Manea et al. (2004) shows, an earlier kink, where there is a temporary steepening of the slab before it flattens out again, has also been postulated, mainly in order to explain the positions of all of the earthquakes that trace out the position of the main thrust.

Cross section showing the inferred geometry of the subducted Cocos plate beneath southern Mexico, showing mostly shallow subduction with a steeper kink about 100 km behind the subduction zone, and slab steep deepening after 250-300 km. Red and yellow circle shows the approximate location of Sunday's earthquake. Grey arrows show inferred bending of the slab. Source: Manea et al., 2004.

Where does last Sunday’s earthquake fit into this geometry? It is located on the second shallowly dipping section of the subducted Cocos plate, about half way between the first steep section and the end of shallow subduction. This is a plausible location for an extensional earthquake; the slab will be starting to get stretched as the plate ahead of it is pulled down more steeply into the mantle. As the figure above shows, there were four other large earthquakes associated with the second bend in the plate; after a very brief search, I found the record of the 1999 6.9, which also has an extensional focal mechanism. So it seems reasonable to interpret last Sundays earthquake as a plate bending event, found where the Cocos plate is starting to transition from shallow to steeper subduction.

Of course, that still leaves us with a mystery: why is the dip of the subducting slab initially so shallow? If you look back to the map in the first figure, you’ll notice that whilst the crust being subducted beneath the west coast of Mexico belongs to the Cocos plate, you don’t have to travel much farther offshore before you cross another plate boundary – this time, a mid-ocean ridge – and find yourself on the Pacific plate. Although the ridge is still active and producing new crust, it is doing so at a slower pace than the Cocos plate is being subducted beneath Mexico. So over time, the ridge is getting closer and closer to the trench, and closer and closer to being subducted itself; a little further to the north the ridge and trench have already collided with each other.

One consequence of this is that the crust being subducted beneath Mexico is not very old; after cooling and solidifying from hot, molten magma at the ridge, it lingers on the surface for just a few million years after being created before it gets pushed back down into the mantle again. And if you’re subducting unsually young crust, that means you’re also subducting unusually warm crust: the way that oceanic crust is created means that it starts hot and gradually cools over tens of millions of years. Because hot, young crust is more buoyant – and therefore harder to subduct – than old, cold, dense crust, this might explain the shallow dip. The additional buoyancy of young Cocos crust makes it want to stay near the surface for a little bit longer, leading to shallow subduction. But eventually the slab is called back to a more proper angle for its return back to where it came from, even if it registers a seismic protest as it does so.