Eruptive Mechanisms in the Minoan Eruption: Evidence from Pumice Vesicularity

Measurements of the density/vesicularity of 30-54 juvenile clasts at each of 61 levels within the deposits at Phira (phases 1 to 3) and Alonaki quarries (phases 3 to 4) yield information on the highly vesiculated state of the magma when fragmented at various stages of the eruption. High clast vesicularities show, with the partial exception of phase 3, that magma:water interaction occurred at levels such that the magma had vesiculated to the point where fragmentation was inevitable (or had already begun) due to the bursting of bubbles. In phase 3 the same is largely true, but a minor population of denser clasts and common breadcrusted pumices implies that some fragmentation by magma:water mixing occurred while the magma was still expanding. Any interaction during the Minoan eruption was occurring at very shallow levels and water never penetrated very deeply into the vent system.

INTRODUCTION

The Bronze Age Minoan eruption generated a complex sequence of deposits that reflect a variety of eruptive styles and secondary fluvial reworking (Bond and Sparks 1976; Pichler and Friedrich 1978; Heiken and McCoy 1984; Sparks and Wilson 1990). The Minoan deposits have been divided into five main units, four of which represent primary volcanic phases. The fifth phase, of fluvial erosion and deposition, is not considered further in this paper. Vertical changes in the pyroclastic succession record variations in vent position(s) and dimensions, emplacement temperatures, discharge rates and the degree of magma:water interaction. Most of these changes can be readily inferred from the combination of field evidence and grain-size data, though many aspects of the emplacement mechanisms and eruption evolution remain controversial. The deposits which formed during the four phases of the Minoan event are as follows (see Sparks and Wilson 1990, for further discussion):

     Phase 1:      A 'dry' plinian fall deposit of modest volume. A thin, poorly sorted, fine-ash rich bed (the phreatomagmatic break) towards its close marks the onset of significant magma:water interaction.

     Phase 2:      Well-bedded, cross-stratified pyroclastic surge deposits displaying spectacular dune bedforms, generated by vigorous phreatomagmatic activity.

     Phase 3:      Deposits of low temperature pyroclastic flows, some of which contained liquid water when they came to rest, generated by vigorous phreatomagmatic activity.

     Phase 4:      A conventional ignimbrite generated by high temperature pyroclastic flows, with no evidence for magma:water interaction.

Field observations (Bond and Sparks 1976; Pichler and Friedrich 1978; Heiken and McCoy 1984) provide ample evidence for the dry emplacement of phases 1 and 4, and cool to wet (i.e. < 100° C at some horizons) emplacement of phases 2 and 3 (cf. Downey and Tarling 1984). However, evidence (below) from sections at Phira and Alonaki quarries emphasizes that there are deposits of intermediate characteristics between the materials considered typical of the four phases.

     Pyroclastic eruption styles:      The diversity of the Minoan deposits has led us to investigate how the relative roles of magmatic volatiles and external water in vesiculating and fragmenting the magma varied during the eruption. The role of magmatic volatiles in explosive volcanism is largely inferred from the vesicularity of juvenile clasts, the clasts preserving a record of the level(s) of vesiculation present when fragmentation interrupted the degassing history of the magma (Houghton and Wilson 1989). With high viscosity magmas, like that generated during the Minoan eruption, two end-member causes of fragmentation are important. Fragmentation by vesiculation (caused by internal, magmatic volatiles) during 'dry' eruptions generates a uniform population of highly to extremely vesicular clasts (typically with vesicularities of 70-85%). Fragmentation by violent mixing of the magma with external water during 'wet' phreatomagmatic activity generates a variety of clast-vesicularity populations, depending on the timing of fragmentation relative to vesiculation, and the heterogeneity or otherwise of vesiculation within the quenched parcel of magma. In 'wet' eruptions we have identified two major groups of deposits (Houghton and Wilson 1989). The first is where magma:water interaction and fragmentation occur at or near the peak of vesiculation so that an expanded and possibly fragmented magma is involved. Such deposits have clast-vesicularity populations dominated (as in dry eruptions) by highly vesicular clasts, though the range of vesicularities is broader than in dry eruptions. The second group is where magma:water interaction occurs well before or well after the peak of vesiculation (i.e. without significant involvement of internal magmatic volatiles). Such deposits have very wide ranges of clast vesicularities, often ranging from 0- > 60%, vesicles (see Houghton and Wilson 1989 for examples).

     Strategy and methods:      The Minoan deposits present unequivocal evidence for 'dry' activity during phases 1 and 4 and large-scale 'wet' or phreatomagmatic activity during phases 2 and 3. We have measured the densities and hence vesicularities of clast populations in the 16-32 mm size fraction from various levels in Minoan deposits to assess the relative timing of vesiculation and fragmentation at different stages during the eruption. Heiken and McCoy (1984) give generalized vesicularity values for pumices in the four eruptive phases, but no information on the class sizes and numbers used, or on their sampling strategy. Druitt et al. (1989) have recorded three types of juvenile material in Minoan deposits: an almost-aphyric pumice (overwhelmingly dominant), crystal-rich pumice and quenched mafic blebs. The density/vesicularity data presented here were obtained only from the almost-aphyric pumices. Between 30 and 54 clasts were collected from narrow stratigraphic intervals at various heights through the deposits (Fig. 2-4).

Techniques used to measure clast densities are detailed in Houghton and Wilson (1989). The clasts were oven-dried, cooled, sealed with a silicone waterproofing spray, and then weighed in air and water to determine specific gravities (and thus densities) by Archimedes' principle. Vesicularities were then calculated assuming a dense-rock-equivalent (particle) density of 2500 kg m^-3. From the raw data we present in Fig. 2-4  the arithmetic mean (hereafter referred to as the vesicularity index), total range and standard deviation of density/vesicularity values, together with the average values of the three lightest and densest clasts, at each level in the deposit.

FIELD OBSERVATIONS

Two stratigraphic sections were measured, the first through deposits of phases 1, 2 and part of 3 in the cutting on the access road at the north end of Phira quarry, and the second through deposits of phases 3 and 4 in the upper part of the Alonaki quarry.

     Phira quarry:      At Phira quarry, a cutting on the access road exposes deposits of phases 1 to 3. Phase 1 (Fig. 2) is represented by a lower 5.3 m thick plinian fall deposit, an intervening series of poorly sorted fine-ash rich beds termed the phreatomagmatic break (Heiken and McCoy 1984) and an upper, thin plinian fall bed. A thin initial phreatomagmatic ash seen elsewhere at the base of the lower plinian fall (Heiken and McCoy 1984; 1990) is absent here. The main plinian fall is very vaguely bedded, with a weak inverse grading, the most noticeable feature being an upwards increase in the contents of foreign lithics and juvenile mafic clasts 4 m above the base. The phreatomagmatic break deposits are generally rich in fine ash, contain accretionary lapilli and vesicular matrices in some beds, and have rounded pumice clasts. These features are generally accepted as resulting from the first onset of significant magma:water interaction in the Minoan eruption, though the presence of the overlying thin, plinian-style fall unit shows that phreatomagmatic activity was not sustained.

Phase 2 (Fig. 3) consists of a bedded, fine-ash rich sequence 7.9 m thick (including beds transitional into phase 3) which is interpreted as the deposits of powerful pyroclastic surges driven by magma:water interaction. The sequence can be divided into three sub-units on the presence of horizons with common ballistic lithics, and these sub-units can be further subdivided into bedform packages. The lower 5.85 m of phase 2 deposits are often low-angle cross stratified or pinch-and-swell bedded, and have variable contents of very fine ash, while other beds are fines-poor and parallel bedded and may represent fall units. The upper 2.05 m is cross-bedded at the base but becomes more massive upwards. This transitional material is demarcated from the rest of phase 2 deposits by the presence of large amounts of a cohesive, often vesicular fine-ash rich matrix and rare breadcrusted pumices, and from phase 3 deposits by its more bedded nature and paucity of > 10 cm lithics.

Phase 3 deposits make up make up most of the section (Fig. 3). They are divisible into flow units, typically 1-2 m thick, on the basis of lithic sizes and contents. Lithic fragments are more abundant than in the underlying deposits and are dominantly black, glassy dacite (Heiken and McCoy 1984). They can exceed 1 m in length but, apart from along the phase 2/3 contact, almost never show signs of ballistic emplacement. Breadcrusted pumices are common. The matrices in phase 3 flow units are variable in texture from loose and rich in fine to coarse ash, to cohesive, vesicular and rich in very fine ash. These variations not only occur between but also within single flow units. One flow unit (29.7-33.2 m in section, Fig. 3) contains two distinct textures, one noticeably richer in lithics. Although it is accepted that the phase 3 deposits were generated by large-scale, violent phreatomagmatic activity, details of the emplacement mechanisms of these deposits remain controversial.

     Alonaki quarry:      The back wall of Alonaki quarry exposes the change from typical phase 3 material, via transitional units, into thin (1.5 m) veneer deposits of phase 4 ignimbrite (Fig. 4). The onset of phase 4 is taken at 2.1 m in the section where the very-fine-ash content and cohesion of the flow unit matrices start to diminish and lithic lithologies become much more diverse. However, flow units above this level are locally cohesive, vesicular and contain bread crusted pumices indicating that 'wet' depositional and thus (by inference) eruptive conditions continued for some time. The base of true phase 4 material is placed at 12.8 m in the section where the flow unit matrices become sandy and non-cohesive, and any traces of vesicularity or 'muddiness' disappear. Phase 4 here is represented by 4.2 m of lithic-rich, fines-poor material overlain by up to 1.5 m of sandy ignimbrite interpreted as a veneer deposit. Palaeomagnetic data from coastal sites suggest emplacement temperatures of > 500° C for the phase 4 ignimbrites (Wright 1978), implying that the influence of water was insignificant at this time.

VESICULARITY DATA AND INTERPRETATION

The Minoan samples show a remarkable uniformity in pumice vesicularity despite the obvious contrasts in eruptive styles. Some subtle variations are present, especially vesicularity of the three densest clasts (V_3H) but, without exception, the bulk of the magma fragmenting at any instant was expanded to a state close to the theoretical limits for vesicularity proposed by Sparks (1978).

All samples are dominated by highly to extremely vesicular pumice and have vesicularity indices between 73 and 82% regardless of any putative magma:water ratios during the eruption. We conclude that at every stage during the Minoan eruption, magma fragmentation could have been initiated by vesiculation alone. The vesicularity data imply that the role of external water was confmed to very shallow levels during the entire Minoan eruption.

The phase 1 plinian deposits were preceded by a fine-grained pumiceous phreatomagmatic ash, suggesting some limited initial interaction of Minoan magma and water. However, the plinian deposits themselves have a very narrow range of vesicularity typical of 'dry' or magmatic fragmentation. There is an upwards decrease in vesicularity index through the main plinian fall (Fig. 2, 5), which appears to accompany an increase in vesicle sizes. The change at 4.0 m (Fig. 2; Fig. 5, sets b and c) also accompanies an increased content of lithics and mafic juvenile clasts. There is no change in pumice vesicularity into the phreatomagmatic break which contains pumice identical in vesicularity to that in the immediately underlying plinian material (Fig. 2, 5). The change to a vesicularity distribution similar to that in the overlying phase 2 deposits actually occurs between the phreatomagmatic break and the topmost plinian band of phase 1.

The phase 2 deposits contain very vesicular pumices but, relative to phase 1 deposits, a 'tail' of slightly less vesicular pumice is present (Fig. 5) and vesicularity indices are slightly denser than in the upper parts of the phase 1 main plinian fall. The transition from phase 2 to phase 3, which marks a major change in depositional styles, is not reflected in any changes of vesicularity indices or in V_3L values (Fig. 3). However, V_3H values and hence standard deviations are higher in phase 3 deposits (Fig. 3).

The phase 4 deposits represent a return to drier eruptive conditions, but only subtle changes in vesicularity characteristics are seen (Fig. 4, 5). These deposits have slightly higher vesicularity indices than phases 2 and 3, but a 'tail' of denser clasts is still present. However, the bulk of the phase 4 ignimbrite seen at lower elevations is probably equivalent only to the topmost veneer deposit at Alonaki, a sample from which is similar to many of the phase 1 plinian samples and lacks the denser clast 'tail' (Fig. 4).

CONCLUSIONS

Field and palaeomagnetic evidence shows that the Minoan deposits have a complex history of emplacement temperatures, from < 100° C during parts of phase 3 to > 500° C in phase 4. These variations are taken to reflect varying degrees of magma:water interaction in the vent area. In general, 16-32 mm clast vesicularities in Minoan deposits are high and uniform, regardless of stratigraphic position, implying that any phreatomagmatic activity took place at levels shallow enough that vesiculation had proceeded to the point where fragmentation was imminent or had already occurred. The only (partial) exception to this is in deposits of phase 3 and its transition from phase 2, where some denser clasts occur and breadcrusted pumices are found, implying that a minor proportion of the magma came into contact with water prior to the peak of vesiculation.

Clast vesicularity data thus provide further constraints on eruptive styles during the Minoan event. The data imply that magma:water interaction occurred at very shallow levels and that water never penetrated deeply into the vent system. If data on pre-eruption magma volatile contents were available, then the maximum depths of magma:water interaction could be estimated from degassing and bubble growth kinetics. We suspect that such calculaticns would show that magma:water interaction occurred within a few hundred metres of the surface.

----------------------------------------------

------------------------------------------