Degassing and magma mixing during the eruption of Surtsey Volcano (Iceland, 1963–1967): the signatures of a dynamic and discrete rift propagation event

Abstract

The eruption of Surtsey (Iceland, 1963–1967) is a rare example of observed volcanism at a propagating rift tip, where Iceland’s Eastern Volcanic Zone is advancing southwesterly toward the Reykjanes Ridge. We use olivine-hosted melt inclusions, embayments, and matrix glasses to investigate major element, trace element, and volatile characteristics of Surtsey’s magmatic system, and to parameterize decompression models that describe degassing of C-H-S-O species during this well-studied eruption. Major elements show that the inclusions represent heterogeneous melts with a range of compositions similar to those of Icelandic lavas. Trace elements discriminate between two groups of inclusions: the majority are compositionally similar to Eastern Volcanic Zone alkalic basalts, and the remainder are notably more depleted in incompatible trace elements and trend toward fields defined by tholeiites on and around Iceland. Strongly correlated Cl and Nb for all inclusions indicates that seawater did not affect Surtsey’s volatile budget, despite having clearly affected its eruption style. Both types of inclusions are volatile rich (≤1.47 wt.% H2O; ≤2534 ppm CO2 as measured or ≤5286 ppm once corrected for vapor bubbles and post-entrapment crystallization). They are more hydrous than primary melts in the onshore Eastern Volcanic Zone and have parental CO2/Nb (≤590) at the upper end of regional estimates. Saturation pressures calculated from corrected CO2 values indicate that melts in both inclusion types initially crystallized at similar depths in the upper mantle (17–21 km), and then partially crystallized during ascent through the lower crust (7–13 km). Historical data show that gases emitted on Surtsey shifted from being relatively reduced in 1964/1965 to more oxidized in 1967, after a protracted period of effusive activity and resurgence of pyroclastic activity from satellite vents. Closed-system degassing models predict the compositions and redox states of the 1964/1965 gases extremely well, but cannot account for the oxidized gases emitted in 1967, which may have been contaminated by ambient air in a system that was opening as the eruption waned. Surtsey’s pyroclastic resurgence can be explained by recharge from ephemeral and compositionally heterogeneous magma bodies, tapped from possibly as deep as the mantle-crust boundary, in a process consistent with the progressively increasing interconnection between magma bodies that is typical at propagating rift tips. The eruption of Surtsey therefore shows that magma system evolution at rift tips can occur in dynamic and discrete events, with influx of new magma having explosive consequences.