The subglacial 1918 eruption of Katla was one of Iceland’s most powerful of the 20th century, producing ~five times the total eruptive volume (DRE) of both
Eyjafjallajökull 2010 and Grímsvötn 2011. However, little is known of the factors that made it such an explosive eruption e.g. was fragmentation fuelled by meltwater interaction or magmatic volatiles?

We have collected both jökulhlaup and airfall tephra from the 1918 Katla eruption. In both deposits, stratigraphy was observed, which has allows us to evaluate different periods during the eruption.

The matrix glass water content, measured using Fourier transform infrared spectroscopy (FTIR), was used to infer loading pressure. Jökulhlaup samples suggest quenching pressures of 0.4 to 1.2 MPa. Although this suggests some loading, it is considerably lower than the 3.6 MPa which is the inferred glacial load, based on an ice thickness of 400 m. There is also evidence of clast welding which suggests that post-fragmentation quenching was not instantaneous and may indicate that fragmentation occurred within the conduit. The H2O content of the airfall tephra is consistent with degassing under atmospheric conditions.

Total volatile contents measured using Thermogravimetric analysis (TGA) suggest that degassing became more efficient as the eruption progressed .

All clasts have a high bubble number density. Vesicles are often spherical with some evidence of coalescence. However, both vesicle size and microlite contents are varied suggesting a variety of cooling rates.

Hotstage experiments, performed at eruptive temperatures, indicate bubble growth rates in the order of 0.8-1.6 µm s-1 . Based on models of pyroclasts cooling within meltwater, this suggests that there would be insufficient time for significant post-fragmentation bubble growth within the jökulhlaup samples. This agrees with vesicle textures as there is often no correlation between bubble size and position within the clast. However, some clasts within the airfall deposit seem to have rapidly quenched margins and significantly larger vesicles in the
clast centre consistent with continued post-fragmentation degassing, suggestive of slower cooling rates which may indicate little or no interaction with meltwater.

Future work involves quantification of bubble textures, further volatile and geochemical analysis and evaluation of clast morphologies and grain size
distributions to infer the degree of water interaction.