Mush Development and Disaggregation in Basaltic Plumbing Systems: Evidence from Large Icelandic Fissure Eruptions

Recent advances in the study of layered intrusions have focussed on the properties of mushy layers that develop at magma chamber margins. However, relatively little effort has been made to integrate the plutonic and volcanic records of basaltic plumbing systems by recognising the signature of mush in erupted products. Furthermore, little is understood about the relationship between the development or destabilisation of mush layers and eruption triggering. In order to investigate magmatic processes in the precursory phases of the Laki Fires eruption we combined detailed petrological and geochemical study of whole-rock samples, tephra glasses and phenocryst phases with simple petrographic measurements. By novel treatment of the data, we find that the Laki Fires whole-rock lava and tephra compositions are controlled by variable incorporation of crystal mush prior to eruption. The Laki Fires fissure eruption took place in AD 1783-4 in the Eastern Volcanic Zone of Iceland. This eruption, with a total erupted volume of ∼ 15 ; km3, is the most recent in a series of large fissure eruptions in Iceland. Lava samples contain phenocrysts of plagioclase (An_72--92 cores), olivine (Fo_66--86 cores) and clinopyroxene (Mg#_66--80 cores). Point counting of 50 lava and tephra samples indicated that these phenocrysts are present with a plag:ol:cpx ratio of 56:14:30 and that porphyritic samples contain a maximum of 35% phenocrysts by mass. The whole-rock samples exhibit modest but statistically significant natural variation in major and trace element compositions. For example, MgO contents vary from 5.64 to 5.88 ; wt% and the range of Zr contents is 174--192 ppm. Strong correlations exist between the major and trace element compositions of the whole-rock samples and the proportion of phenocrysts present in each sample, with decreases in the concentration of incompatible elements with increasing phenocryst content. Simple accumulation of phenocrysts in a carrier liquid generates perfect correlations with negative slopes on plots of whole-rock incompatible element concentration against phenocryst content, with the intercept at zero crystal content being the carrier liquid composition and the intercept at a phenocryst fraction of 1 being the average phenocryst composition. However, the Laki Fires samples do not sit on these simple crystal accumulation lines because the extrapolation of the observed trends to a solid fraction of 1 does not correspond to the observed phenocryst composition. This observation can be accounted for by a model where the Laki Fires carrier liquid mixes with a mush containing crystals and an evolved mush liquid. We used the relationship between the crystal content and whole-rock chemistry and the composition of phenocryst rims to constrain both the composition of the mush liquid and the relative proportions of liquid and crystal in the mush. These constraints can be used to estimate the physical properties of the mush and are therefore crucial to the development of fluid dynamical models of magma chamber processes prior to large basaltic fissure eruptions.