Magmatic evolution of Krafla, N.E. Iceland

Abstract

Krafla is an active central volcano in the, NE axial rift zone of Iceland associated with a fissure swarm which trends NNE-SSW. Lava/hyaloclastitc samples were collected from the volcanic system, covering the last 4 interglacial and 3 glacial periods. A stratigraphic framework for the volcanic system has been established by use of the Hekla tephra layers and the distinctive volcanic products of glacial and interglacial periods. Recent volcanic activity has shown that there is a shallow magma reservoir beneath the central volcano, which supplies magma laterally to the fissure swarms forming dykes and to the surface directly for eruption.

The Krafla rocks contain the following phenocryst assemblages ol + plag, ol + plag + cpx, plag + cpx ± ol, plag + cpx + opx + FeTi oxides, plag + FeTi oxides cpx ± fayalitic ol. These distinct assemblage suggest that the Krafla suite evolves predominantly by fractional crystanation. They are also similar to the assemblages found in other mid-ocean ridge basalt (MORB) suites.Most samples, excluding rhyolites,contain plagioclase xenocrysts,implying that there has been at least 2 stages of magma mixing prior to eruption The absence of plagioclase xenocrysts in the rhyolites suggests that they formed in isolation from the most primitive magmas with or without crustal assimilation.

Fractional crystallisation using the observed phenocryst assemblage explains many aspects of the 4 major-element compositions of the suite. Least-squares modelling confirm this observation, although it also suggests that the rhyolites contain higher concentrations of K₂O than would. be predicted by closed-system fractional crystallisation. This discrepancy, may be, explained by open system fractionation and/or by crustal assimilation. Numerical modelling of phase equilibria suggests that fractional crystallisation occurs over a range of pressures from 1-3.5 kbar. As was suspected, magma mixing also'appears to be a significant process in the Krafla f system and may explain sorne of the scatter on major-element variation diagrams. The Krafla suite is more differentiated than the majority of MORB suites, consistent with the development of long-lived magma reservoirs. The suite also has higher average values of total iron oxide and lower average values of Na₂O (for a given MgO content), which are explicable by larger degrees of melting at'higher than average pressures than for the majority of MORBs. i, Trace-element compositions also confirm'the role of fractional crystallisation. Although the trace-element data also show that the highly incompatible trace elements (e. g. Th, U, Rb) show have high concentrations in the rhyolites like K₂O. As mentioned above, this enrichment in the most, incompatible elements may be explained by either open system effects and/or by crustal assimilation. However, the temporal variation in rock'composition shows no evidence for enrichment in the more incompatible elements when compared with the less incompatible elements, effectively reducing the role of open system fractionation. This suggests that the enrichment in'inc'ompatible elements occurs through crustal assimilation of possibly basaltic wall-rock which has undergone relatively small degrees of melting. Ratios of incompatible trace elements also suggest that the Krafla basalts are distinctive from most MORBs in that they appear to be derived from a less incompatible element enriched mantle source.

The major- and the trace-element compositions of the most primitive Krafla basalts show evidence for variable degrees of melting of mantle or possibly variations of the source. In particular, the major-element concentrations, when compared with data from experimental studies, suggest that the basalts derived from the smallest degrees of melting are also produced at the highest pressures. The variation in the incompatible trace elements appears to be "coupled" to that seen in the major elements. The least-differentiated Krafla compositions are also compared with melt compositions predicted from the parameterisation of melting experiments (McKenzie & Bickle 1988). After correcting for the effects of fractional crystallisation, the Krafla melts are most consistent with model compositions produced by a value of 1480°C for the mantle potential temperature. Discrepancies between model compositions and the Krafla compositions may possibly be explained by the dynamic effects of a mantle plume on the melting processes.

The Krafla lavas show anomalously low δ¹⁸O values compared with the majority of MORBs. The δ¹⁸O values of the lavas correlate positively with the MgO content. This correlation is consistent with crustal assimilation of low-¹⁸O hydrothemidly-altered country-rock. The assimilation process appears to be coupled to fractional crystaffisation, although the linearity of the δ¹⁸Ovs MgO plot appears to confirm that magma mixing is also operating. The (²³⁰Th/²³²Th) values of postglacial rocks also correlate positively with MgO content, consistent with their origin by the same crustal assimilation process. The 0- and Th-isotope ratios may be modeHed by an assimilation with fractional crystallisation (AFC) model, providing the δ¹⁸O value of the assimilant is assumed. An assimilant produced at the top of the magma reservoir is likely to have a 8180 value of about -100loo. If this value of 8180 is used for the assimilant, then a single-stage AFC model requires that the assimilant has to contain between 3-5 ppm, Th and r=0.15-0.20. A 2-stage model, however, gives a better fit with the O-Th isotopic data, with an assimilant containing about I ppm Th in the first stage of differentiation and 7 ppm in the second stage. Both models can generate the high values of the most incompatible trace elements in the rhyolites.