Lateral vs vertical propagation of dikes through analogue modelling

Abstract

The interest on the mechanisms governing dike propagation has been renewed by recent diking events showing lateral propagation along rift zones for tens of kilometers. Such laterally propagating dikes may feed eruptions far from their magma chamber, thus representing volcanic hazard in densely populated areas far from central volcanoes. It is therefore crucial to understand what mechanisms drive dike propagation and arrest in order to be eventually able to constrain the possible location of surface vents. In this regard, it is known that dikes may propagate both laterally and/or vertically at both a local (central volcanoes) and regional scale (rift systems), depending on several factors (such as regional stress, crustal layering, topography, buoyancy, magma solidification and gas exsolution, preexisting structures in the host rock, magma influx rate). However, so far these factors have been usually studied independently of each other. As in nature these factors may act contemporaneously, it is crucial to assess their relative importance by establishing a hierarchy. The goal of this thesis is to address this latter issue by defining an hierarchy between a selection of these factors (crustal layering, topography, density ratio between host rock and magma, magma influx rate) through analogue models focusing on the conditions favoring lateral propagation. In the analogue experiments, each of the studied parameters has been varied systematically, fixing all the other variables, comparing semi-quantitatively the variations on dike shape and velocity. Two different sets of experiments with vertical (Set 1) and lateral (Set 2) injection have been performed. The results indicate that rigidity layering (i.e. a stiff layer overlying a weak one) and topography are the most efficient parameters that favor lateral propagation, while density layering, density ratio and magma influx rate show a subordinate role. Moreover, the observations made during the experiments of dike propagation testing the joint effect of rigidity layering and topography (i.e. downslope lateral dike propagation and arrest in front of an external relief) show interesting similarities with the 2014 Bardarbunga (Iceland) diking event, suggesting that our results should be taken into account to explain the lateral propagation observed in other recent diking events. In nature, rigidity inversions (i.e. stiff layers above weak ones) may occur frequently. These inversions are usually due to alternating soft scoria layers and stiff lava flows (as in Afar), or to thermal weakening, which decreases host rock rigidity at higher depths (as observed in Iceland, Hawaii and Canary Islands). Such rigidity contrasts may be one order of magnitude higher than the contrasts imposed in the models, thus encouraging natural dikes to propagate laterally at the interface with the stiffer zones as observed in the models. This approach to the investigation of the factors controlling dike propagation (i.e. systematic variations of the parameters and hierarchy definition) should be followed in future works adding the remaining processes that influence dike propagation (such as pre-existing structures in the host rocks, magma solidification, regional tectonic stresses, gas exsolution) not considered in this study.