Mechanical Interactions in South Iceland Pleistocene Basalt-Hyaloclastite Sequences

In southern Iceland, a series of Pleistocene eruptive sequences, likely derived from proto-Laki eruptions, displays unusual basalt-hyaloclastite interaction (Bergh and Sigvaldason, 1991). These sequences are exposed over an E-W distance of 75 km in paleo-seacliffs and display alternating layers of basalt lava and basaltic hyaloclastite, with lava thicknesses reaching up to 50m and overlying massive hyaloclastite thicknesses reaching over 100m. The lavas are composed of a "standard" sequence of columnar lava overlain by cube-jointed lava, with columnar lava absent in some areas. Throughout the exposed cliff sections, cube-jointed lava injects into and is disaggregated within the overlying hyaloclastite, yielding pillows, pods, flame structures, and dikes. The geometry and scale of these structures vary dramatically by location. Bergh and Sigvaldason (1991) proposed simultaneous subaqueous eruption of both lava and hyaloclastite with entrainment of lava accompanying mass flow. Emplacement of lava flows preceding hyaloclastite, rather than as sills, is supported by a relative lack of vesiculation in the lava, indicating considerable post-eruption degassing. We propose that the lavas were erupted from an ice-free, subaerial portion of the fissure. As the eruption propagated from subaerial eruption to subglacial, the recently-erupted lava flows were quickly buried by hyaloclastite deposited from a jokulhlaup generated by eruption-induced glacial melting. Heat was introduced from contact with the lava, still molten beneath a thin crust. Fracturing of the crust led to a volume expansion of the water in the hyaloclastite, leading to a pressure field that locally weakened the hyaloclastite, allowing injection to occur. Freezing of lava margins once the lava propagated upward into the hyaloclastite allowed the liquid lava to continue vertical propagation. Rheological variation in hyaloclastite related to water content, temperature, and material sorting likely contribute to the variation in style of injection and disaggregation. Approximately 250 EMP and LA-ICPMS analyses of fresh glass from hyaloclastite samples from 3 representative locations revealed uniform compositions, consistent with a common source. The major element composition of about 0.3 wt% K2O, 2.5 wt% TiO2, 7 wt% MgO, 2.5wt% Na2O, and 13 wt% FeO is similar to modern Laki elemental values (Lacasse et al., 1998; Metrich et al. 1991). In contrast, 35 analyses of glass from lava at the base of the section in its easternmost exposure display elemental signatures that are markedly different from the hyaloclastites, including the one that overlies it. Compared to the hyaloclastite glasses, this glass displays distinctively higher SiO2, FeO, and Na2O, 2x to 4x higher REE, TiO2, K2O, and P2O5, and lower MgO and Al2O3. This puzzling relationship might be a consequence of one of the following: (a) The lava was sourced from fractionated melt at the top of a magma chamber and erupted subaerially; the less-evolved magma that immediately followed erupted subglacially, from a slightly different location, formed the hyaloclastite, and was transported over the recent lava; (b) The lava and hyaloclastite were sourced simultaneously from a laterally variable magma chamber; or (c), The lava and the hyaloclastite were sourced from identical magma (though at separate locations along the fissure), but the hyaloclastite melt was immediately quenched to preserve the initial liquid composition, whereas the lava fractionated during transit from the eruption site to yield