Earthquake prediction has not had a good record, but it seems that vastly larger tectonic processes are now becoming the subject of risk analysis (Nikolaeva, K. et al. 2011. Numerical analysis of subduction initiation risk along the Atlantic American passive margins. Geology, v. 39, p. 463-466). The Swiss, Russian and Portuguese authors focus on the old (Jurassic ~170 Ma) and presumably cold oceanic lithosphere on the western flank of the Atlantic, against both the North and South American continents. Increased density with ageing imparts a potential downwards force, but that has to overcome resistance to plate failure at passive margins. The dominance of upper continental lithosphere by rheologically weak quartz tends to make it more likely to fail than more or less quartz-free oceanic lithosphere. So, if subduction at a passive continental margin is to take place, then where and when it begins depends on the nature of the abutting continental lithosphere. That on the Atlantic’s western flank varies a lot, ranging from 75-150 km thick. Consequently the temperature at the Moho, the junction between continental lithosphere and weaker asthenosphere, varies too. The loading by marginal sedimentation also plays a role, as do continent-wide forces associated with far-distant mountain ranges, such as the Western Cordillera and Andes, and the forces from opposed sea-floor spreading from the Juan de Fuca and East Pacific systems that affect the whole of western South America, most of Central America and the far NW of North America.
Analysing all pertinent forces acting along 9 lines of section through both North and South America, the authors’ focus fell on the relatively thin continental lithosphere of the Atlantic margin of South America. It is at its thinnest along the southernmost part of the margin adjacent to Brazil, where the Moho temperature reaches as high as 735°C: the weakest link in the American continental lithosphere, where there is seismicity and also indications of igneous activity. The modelling suggests that incipient deformation may begin off southern Brazil within 4 Ma to form a zone of overthrusting, eventually evolving towards failure of the ocean-continent interface and the start of proper subduction in the succeeding 20 Ma. Other stretches of the eastern Americas are deemed safe from subduction for considerably longer by virtue of their greater thickness, lower Moho temperatures and thus higher strength. It is an interesting situation because, insofar as I understand plate tectonics, extensional or compressional failure needed to generate plate boundaries must also propagate from the weak spots that first fail; plate boundaries are lines not points. If that does not happen, then the very strength of the overwhelming longer continent-ocean interface will surely prevent subduction at a single, albeit weak link.
Paper PDF at http://xa.yimg.com/kq/groups/13231164/1842350625/name/Geology-2011-Nikolaeva-463-6.pdf