DYKEHAZARDS · When do injected dykes reach the surface to feed eruptions in a volcano? Field observations, numerical modelling, and hazard implications

Final Activity Report Summary - DYKEHAZARDS (When do injected dykes reach the surface to feed eruptions in a volcano? Field observations, numerical modelling, and hazard implications)

Most dykes injected from magma chambers become arrested at depth without giving rise to an eruption. Dyke propagation can be registered at surface as an unrest period, with seismicity, deformation and/or anomalous emission of gases. The aims of this project were to improve our methods to: (1) infer correctly if an injected dyke is likely to be arrested or is a potential feeder, and (2) to forecast how an eruptive fissure is likely to propagate. We used detailed field observations and data on the mechanical properties of volcanic rocks to develop realistic numerical models as to points 1 and 2. The main results of this study relevant to hazard and risk assessment are as follows. (1) Injected dykes only reach the surface (to feed eruptions) when the local stress fields in each layer between the source chamber and the surface favours the propagation of magma-driven extension fractures. If there are layers and / or contacts with unfavourable stress fields, the potential feeder dyke will become arrested so that an eruption is prevented. (2) From point (1) it follows that the likelihood of an eruption during an unrest period with dyke injection can be assessed through stress monitoring, that is, in-situ stress measurements in drill holes in the volcano. Stress-monitoring of this kind is now being proposed for an ICDP project on Campi Flegrei, Italy. (3) Dyke-induced surface deformation can be of various types and depends strongly on mechanical layering and contact properties close to the surface. Thus, while dykes have widely been thought to generate grabens (normal faults) at the surface above their tips, this project reports, for the first time, a clear example of a dyke inducing reverse faulting on an existing graben boundary fault. (4) In the project it is concluded that most dykes propagate as discontinuous 'fingers' which, when reaching the surface produce offset, discontinuous fissure segments. The evolution and linking up of these segments is analysed through numerical models, from which the effusion rate changes during eruptions can be inferred. All these results are of importance for lava flow models and for civil authorities for assessing hazards and risks during unrest periods and eruptions. More specifically, surface deformation induced by propagating dykes depends strongly on mechanical layering. In particular, weak or open contacts at shallow depths can partly control dyke induced surface stresses and associated deformation, making direct inversion (using isotropic half-space models) of surface-deformation data using isotropic obtained during unrest periods to infer dyke dimensions and dyke-tip depth unreliable. Also, some dykes appear to be captured by grabens and the dyke-induced stress may cause reverse faulting (not normal faulting) on the boundary faults, as reported in this project. The detailed geometry of a dyke path, and whether or not it eventually reaches the surface (as a feeder dyke), depends on mechanical contrast between adjacent layers and, in particular, the mechanical properties of the contacts between layers. The present data suggest that a dyke normally propagates as a set of discontinuous segments or 'fingers'. If the dyke propagation continues to the surface, the finger that propagates fastest is the one that first intersects the surface. At the surface the dyke-fed segments of the volcanic fissure may propagate laterally, some eventually coalescing, to form the final volcanic fissure. New numerical models show that faults and caldera walls (ring faults) can direct the volcanic-fissure propagation direction. In particular, dykes propagation paths may be partly controlled (captured) by existing faults, as is supported by field observations. In the project a study was made of a remarkable phonolithic feeder dyke with giant bubbles under the vent deposits. Comparatively large bubbles (vesicles) may be used as one criteria to identify feeder dykes in eroded volcanoes and rift zone, and to assess the proportion of feeders versus non-feeders in eroded dyke swarms. Present data suggest that a dyke normally propagates as a set of discontinuous segments or 'fingers'. If the dyke propagation continues to the surface, the finger that propagates fastest is the one that first intersects the surface. At the surface the dyke-fed segments of the volcanic fissure may propagate laterally, some eventually coalescing, to form the final volcanic fissure. New numerical models show that faults and caldera walls (ring faults) can direct the volcanic-fissure propagation direction. In particular, dykes propagation paths may be partly controlled (captured) by existing faults, as is supported by field observations. In the project a study was made of a remarkable phonolithic feeder dyke with giant bubbles under the vent deposits. Comparatively large bubbles (vesicles) may be used as one criteria to identify feeder dykes in eroded volcanoes and rift zone, and to assess the proportion of feeders versus non-feeders in eroded dyke swarms.

Data: CORDIS, © European Union