Agust Gudmundsson

Dikes supply magma to most volcanic eruptions. Understanding how propagating dikes may, or may not, reach the surface is thus one of the fundamental tasks for volcanology. Many, perhaps most, dike segments injected from magma sources do not reach the surface to feed volcanic eruptions. Instead, the dike segments become arrested (stop their propagation), commonly at or close to contacts between mechanically dissimilar layers/units, at various crustal depths. This means that many and perhaps most volcanic unrest periods with dike injections do not result in eruptions. There are several conditions that make dike arrest likely, but the main one is layering where the layers have contrasting mechanical properties. Such layering means that local stresses are heterogeneous and anisotropic and, therefore, in some layers unfavourable for dike propagation – hence the dike arrest. Here I show that once a dike has formed, however, its very existence tends to make the local stress field along the dike homogeneous (with invariable orientation of principal stresses) and favourable (with dike-parallel orientation of the maximum compressive principal stress) for later dike injections. This means that subsequence dikes may use an earlier dike as a path, either along the margin or the centre of the earlier dike, thereby generating a multiple dike. Because earlier feeder-dikes form potential paths for later-injected dikes to the surface, many volcanic eruptions are fed by multiple dikes. Examples include recent eruptions in the volcanoes Etna (Italy) and Kilauea (Hawaii), and the Icelandic volcanoes Krafla, Hekla, and Fagradalsfjall. Thus, multiple dikes favour dike propagation to the surface; thereby making dike-fed eruptions easier.