The Upper Thera (Minoan) Ash in Deep-Sea Sediments: Distribution and Comparison with Other Ash Layers

The importance of this ash to both archeologists and geologists is apparent in that it has formed the basis for two international congresses on the volcano Thera, stimulated by the work of Marinatos (1939), and Ninkovich and Heezen (1965). The tephra deposit was first recognized by Mellis (1954) in Albatross piston cores, the first extensive oceanographic cruise to utilize a piston corer. Later work by Olausson (1960, 1961), Ninkovich and Heezen (1965, 1967), Keller (1971), Keller and Ninkovich (1972), Ryan (1972), McCoy (1974), Richardson and Ninkovich (1976), Keller et al. (1978) and Watkins et al. (1978) delineated the areal distribution of the upper Thera ash in deep-sea sediments and established its correspondence with the Oberer Bimssteinhorizont (Bo) of Reck (1936) on Santorini Island. Only cores from the Albatross, Vema and Conrad were available for study by Ninkovich and Heezen in 1965. Many additional cores have been taken and studied since their study by the ships Chain, Pillsbury, Atlantis II, Trident and Eastward, all of which have been used here (Figure 2); important new information has also come from studies of the ash layer on land (Bond & Sparks 1975; Friedrich & Pichler 1976; Pichler & Schiering 1977; Keller, this volume; Vitaliano, D.B. and C.J. 1978). These new data allow a more detailed determination of the overall ash distribution both as it was immediately following the eruption and as it is found now. The upper Thera (Minoan) ash is only one of twenty-five volcanic ash layers present within the late Quaternary stratigraphic sequence in the Eastern Mediterranean and Aegean Seas, and comparisons will be made between it and the other tephra layers as an indication of the volcanological, sedimentological, and archeological significance of the Minoan eruption.

TECHNIQUES AND METHODS

• Piston Coring

An important factor in realizing the constraints on interpreting deep-sea sedimentological data is an understanding of the coring process. Basically, a 2.5 to 3 in. (6.4 to 9.6 cm) diameter hollow pipe, 10 to 60 ft. (3 to 18 m) long (depending on the sediment type), is forced into bottom sediments by a heavy weight (1500 - 2000 lbs; 680 - 910 kgs) after free-falling some predetermined distance (12 to 20 ft; 3.6 to 6 m) above bottom, triggered by a bottom-sensing, smaller corer known as the trigger-weight (TW) corer (Fig. 3). There are two fundamental types of corers: gravity corers and piston corers. The piston corer has an internal piston adjusted to remain at the sediment-water interface while the corer penetrates into bottom sediments. The piston prevents excessive compaction of the cored material and considerably reduces internal friction between the cored sediment and the downward-moving pipe thereby significantly increasing the amount of penetration; essentially it is the same principle as a syringe, but with the plunger (piston) theoretically remaining immobile. In practice, however, it is difficult to accurately position the piston considering the elasticity of steel cable in water depths often exceeding a mile or more. Inaccurate positioning of the piston can result in disturbed cores that have been shortened or have poor stratigraphic integrity. Photographic studies of the coring process indicate that up to a meter or more of the uppermost sediment can be disturbed and not sampled due to the impact of the heavy piston corer with the bottom (McCoy & Von Herzen 1971).

The more gently penetrating activity of the TW gravity corer usually recovers this uppermost meter of sediment. Careful comparison of the TW core with the upper portion of the piston core will indicate such disruptions or disturbances and, when necessary, is used to reconstruct the entire sediment column. The use of TW corers was initiated after the Albatross expedition, where lead weights were used as the free-fall triggering mechanism, because up to a meter of sediment was found to be missing from the tops of most of these cores (Emiliani 1955; Olausson 1961). Since the Upper Thera (Minoan) ash occurs within this upper interval, many of the Albatross cores did not sample the tephra deposit.

One important factor remains unknown: the amount of compaction in cores due to the sampling process. There is no method yet to determine if, for example, a 1 cm thick tephra layer in a core represents a 1 cm thick layer in situ. Indications are that cored intervals are within 10% of those in situ but this is largely dependent upon the physical properties (shear strength, water content, etc) of the sediment.

Despite these problems, the piston corer is the only effective sampler capable of obtaining long cores in the upper few meters of deep-sea sediments and has served the oceanographic community well. The operational and design constraints noted here must be kept in mind, however, in interpreting data collected by coring.

• Laboratory Methods

Individual ash layers in the eastern Mediterranean pelagic section can be readily identified by

1. stratigraphic position with respect to sapropel layers;
2. the color of the ash deposit;
3. refractive indices of glass shards;
4. morphology of these shards;
5. accessory minerals; and
6. the chemical composition of the ash.

Sapropel layers are distinctive basin-wide deposits of dark-grey to black organic-rich sediments that reflect periodic bottom-water stagnation due to sea-level changes during Quaternary climatic cycles. Overall color is distinctive for only some of the tephra deposits and can be negated by natural or artificial (such as in core opening) mixing with host sediments. The other criteria represent internal characteristics of the ash related to its chemical composition and thus its volcanic source. Tephra dispersal patterns have been interpreted from areal variations of layer thickness, grain-size variability and glass to crystal ratios, although these can be modified by benthic biological mixing activity and redispersal processes. The combination of a few of these attributes is adequate to define individual tephra layers, delineate their distribution pattern and identify the volcanic source; all have been utilized here.

Refractive indices of glass shards were optically determined by calibrated immersion oils on ten grains or more with appropriate temperature corrections applied whenever necessary; accuracy was within ± 0.001. Carbonates were removed prior to all analyses by mild acid treatment. Samples were studied microscopically using smear slides and grain mounts, with at least one thin-section made from each ash layer. Carbonate-free samples were separated into sand (> 62μ), silt (62-2μ) and clay (< 2μ) fractions by wet-sieving, dispersal with peptizers, and centrifuging following the procedures of Hathaway (1956), weighed, then sands and silts recombined on glass slides for grain mounts. Point counts were made from these mounts of vitric, lithic, and mineral components on 100 - 300 grains; mineralogic determinations utilized both thin-sections and grain mounts. For distinguishing crystalline components in the sub-microscopic size range, small portions of the silt-sized fraction of each tephra were analyzed by standard x-ray diffraction techniques. Wherever possible, ash layer thicknesses and colors were determined on moist cores within hours of being opened and split thereby diminishing problems associated with shrinkage and color changes as cores dried.

The absence of ash in a core taken from an area where ash layers are known to be present from several surrounding cores was not considered significant due to the processes of biologic bioturbation, slumping in areas of high sea-floor relief, and possible coring disruptions. Boundaries of thephra dispersion have been determined by the absence of the tephra in several cores within the same physiographic area.

CHARACTERISTICS AND DISTRIBUTION OF THE ASH LAYER

• Characteristics

The upper Thera (Minoan) ash usually occurs in cores within the upper meter of the sediment column about a third of the distance downsection (depending upon sedimentation rates) between the top of the core (as reconstructed using the TW core) and the upermost sapropel, S1 (sapropel chronology and nomenclature follow that of McCoy 1974 and Cita et al. 1977). It is a light-grey, rhyodacitic, vitric ash often indistinguishable from the similarly-colored muds and biogenic oozes forming the host sediment. Vitric grains are predominantly colorless to light yellow-brown pumice fragments, elongate to equant in shape, often containing microlitic inclusions, with a subordinate number of curved or "Y"-shaped shards. Both types are typical of shard morphological types associated with rhyolitic and dacitic magmatic eruptions (Heiken 1972), and are characteristically present in these proportions in tephra produced by Hellenic arc volcanoes. Glass refractive indices vary between 1.506 and 1.512, with average values of 1.508 to 1.510, reflecting the SiO₂ content of the ash (Figure 4).

The mineral fraction constitutes between 12% (Keller 1971) and 2% of the tephra, the lower amounts typical in distal portions of the submarine deposit. In outcrop on Santorini Island, mineral content forms only about 8% to 5% of the ash (Bond & Sparks 1975; Vitaliano et al. 1978). The mineralogical composition of plagioclase, hypersthene and augite reflect its rhyodacitic composition and derivation from the calcalkaline Thera volcanics (Nicholls 1971; Keller 1971; Richardson & Ninkovich 1976; Keller et al. 1977).

In chemical composition, the deep-sea material (Keller et al. 1978) is closely similar to the chemical composition of the basal pumice (Rosa Bims of Reck 1936) of the Upper Thera ash on Santorini (Pichler & Kussmaul 1972; Vitaliano et al. 1978), and apparently represents material ejected during the initial pumice-fall phase of the eruption as is typical of Plinian-type volcanic eruptions in general (Sparks et al. 1973).

• Present-Day Thickness Distribution

Ash thickness variations in cores indicate extensive post-depositional mixing during the past 3400 - 3500 years since the eruption by slumping and benthic biological mixing processes (Figure 5). Present-day thickness variations mimic bathymetry (Figure 1): thick accumulations follow depth contours outlining the southern Aegean Sea trough north of Crete and west of Rhodes, implying slumping and resedimentation of tephra from steep surrounding slopes into the trough. Thin accumulations south of Santorini may be due to this slumping, in addition to mixing of the ash into host sediments by benthic organisms and dispersal by currents. Within the rugged topography of the Hellenic trench system southeast of the Hellenic arc, a distinct zone of thin ash layer accumulation defines the high-relief areas flanking trenches indicating slumping here as well, a factor emphasized by Olausson (1971). These post-depositional modifications to ash layer thicknesses are also evident in the deviation of the expected exponential decrease of thickness with distance from the eruptive vent (Figure 6).

A few short cores taken south of Santorini in the Aegean Sea appear to have recovered between two to four layers of the tephra. Published descriptions of these, however, suggest much mixing of volcanic particles into host sediments from one thick deposit rather than implying multiple air-fall eruptive phases. If the eruption was typical of other Plinean-type eruptions, as interpreted from the geologic record (Sparks et al. 1973), only the initial eruptive phase of non-welded debris (pumice fall or co-ignimbrite tephra fall) would have resulted in a distinct tephra layer in pelagic sediments.

It is clear that there has been much redeposition of volcanic material in the deep-sea. Even in those cores where mixing processes appear to have been at a minimum and layer thicknesses presumably approximate those of the original deposit, there is still a noticeable amount of shards mixed into the host sediments. It is remarkable in fact that any distinctive tephra layer remains in deep-sea sediments after these dispersal and mixing processes, in combination with possible disruptions resulting from sampling.

• Original Thickness Distribution

In attempting to reconstruct original ash layer isopachs and dispersal patterns, such post-depositional modifications must be considered. Additional criteria have come from: reported volcanic particles occuring in trace amounts in soils and Minoan ruins on Crete where glass shard refractive indices are similar to those measured in the upper Thera ash elsewhere (Boekschoten 1971; Vitaliano & Vitaliano 1974; Pichler & Schiering 1977; Cadogan & Harrison 1978) and it should be pointed out that the low values found by Boekschoten are within the lower limits of those for the ash (Figure 4); from isopachs of the tephra layer on Santorini (Bond and Sparks 1975); and, most significantly, from the eastern tip of Kos where Keller (this volume) has identified one 30 cm thick layer as being the upper Thera ash.

Applying these data and suggested post-depositional changes, the original distribution of the ash and the variation in layer thicknesses within pelagic sediments is depicted in Figure 7. Reported tephra layer thicknesses on land have been corrected for comparison to those measured in cores by applying a 50% reduction factor (Thorarinsson 1971).

A more easterly dispersal pattern is indicated than originally suggested by Ninkovich and Heezen (1965), but is in agreement with data presented by Bond and Sparks (1975) from Santorini, and inferences by Watkins et al. (1978), and Keller (this volume) from Kos. Isopachs from Turkey are difficult to determine. Dispersal by easterly geostrophic winds in the stratosphere is still applicable, as suggested by Ninkovich and Heezen (1965). Along the southern margin of the distribution pattern, there apparently was dispersion of the ash to the south by atmospheric winds (Figure 8) and/or by surface and intermediate-depth oceanic currents (Figure 9), although neither has seriously modified the overall pattern to the extent noted, for example, by Ninkovich and Schackleton (1975) for ash layers in the Panama Basin. There was significant dispersion of pumice by surface currents throughout the eastern Mediterranean Sea at the time of the eruption since deposits of upper Thera pumice have been identified on elevated beaches on Cyprus (Fornaseri, et al. 1975; Åström, this volume), in archeological sites in the Peloponnese (Rapp et al. 1973) and on Milos (Vitaliano & Vitaliano 1978), and implied in strand line deposits in Syria (Van Liere 1961) and possibly Israel (Pfannenstiel 1952), although the latter occurrence is dubious (Keller 1971). Abrasion during movement of this pumice, presumably as pumice rafts, has not left an identifiable ash layer in the deep-sea record.

Such a pattern for primary dispersal of tephra implies that any destructive effects on inhabitants in the southern Aegean Sea region due to the ash fall would have been concentrated to the east of Thera as far as Turkey, rather than to the south on Crete.

Suggested ash accumulations of up to 40 cm on Rhodes, Karpathos, many of the Dodecanese islands like Kos (Keller, this volume), and portions of southwestern Turkey could have resulted in serious damage to settlements lasting for decades, applying the criteria of Thorarinsson (1971). Accumulations on Crete, however, were probably no more than 6 cm at the maximum, with an average closer to about 1.5 cm, thus being no more than a temporary and brief nuisance to settlers. Thus it is not surprising that evidence of the ash fall has been so meager on Crete; future efforts by both geologists and archaeologists would be better directed towards areas like Rhodes and southwestern coastal Turkey.

COMPARISON WITH OTHER ASH LAYERS IN DEEP-SEA SEDIMENTS

Within late Quaternary sediments of the eastern Mediterranean Sea, twenty-five volcanic ash layers have been identified in cores (Keller et al. 1978; McCoy, in press), representing a remarkable history of explosive volcanic activity during the past 240,000 years. The tephrochronology of these deposits, in comparison with the sapropel chronology, oceanic oxygen isotope record determined from planktonic foraminifera, and the planktonic foraminifera zonation of Ericson and Wollin (1968) - the latter two being indicators of climatic changes - are summarized in Figure 10. 

Sapropel layer identification follows the method devised by McCoy (1974) and Cita et al. (1977). Using the scheme of Keller et al. (1978), ash layer nomenclature involves first a letter, which identifies the planktonic foraminifera zone it occurs in, and second a number giving the sequential, downcore stratigraphic position of the ash within that zone. An additional identifier is given to describe either the specific source volcano, where known, or the general petrographic province where the presumed source is indicated according to petrologic and petrochemical data.

Two areas of volcanicity have been the sources for these tephra layers:

1. the Hellenic arc area which has produced calc-alkaline volcanic rocks, and
2. the island arc area of Italy - Aeolian Islands - Sicily - Pantelleria which has produced alkalic, peralkalic, leucitic and calc-alkaline rocks with distinctive chemical differences from those of the Hellenic arc.

This contrast in petrographic provinces results from primary magma variations due to differences in tectonic activity between the two geographic areas (Galanopoulos 1973; Ninkovich & Hays 1971, 1972) and it is because of these contrasting tectonic and petrographic factors that the resulting volcanic debris in deep-sea sediments is characteristically identifiable, particularly by the variation of refractive index values of glass shards which reflect the variation in silica content (Figure 4).

Comparison of the distribution between the upper Thera ash (Figures 7 and 11) and the other ash layers in this marine sequence (Figures 11 through 14) demonstrates the similarity in their east to southeasterly dispersal patterns, with the exception of ash layers V-1 and V-3. These patterns are due to stratospheric and atmospheric wind transport of volcanic material during the past 240,000 years, implying persistant wind directions despite significant climatic changes. Although the effects on dispersal by oceanic currents and post depositional processes are important, aeolian transport has been the dominant process for primarily dispersing volcanic ejecta over the eastern Mediterranean Sea. The reasons for the differing patterns illustrated by ash layers V-1 and V-3 cannot as yet be determined since neither has been adequately sampled in cores.

While Hellenic arc volcanoes have been significant in the production of tephra, most of the ash layers in the deep-sea record have originated from volcanoes in the Calabrian area. This reflects the relative numbers of volcanic centers between the two geographic (tectonic) areas as well as the more explosive nature of Calabrian volcanoes.

Inferences made from the distribution patterns with respect to the inferred difference in eruptive explosivity and the volume of ejected tephra indicate that the ca. 1450 B.C. Thera eruption was not an extraordinarily large eruption in geological terms. The Y-5 ash, also known as the "Ischia" ash, was derived from an enormous eruption in the Neopolitan area about 35,000 years ago and deposited an extensive ash layer (Figure 12) that dwarfs the Minoan eruption. The Vesuvius (Z-1) ash, the X-1, W-1, and V-3 ash layers all appear to represent significantly larger eruptions relative to the Minoan (Z-2) eruption, at least in terms of ash dispersal patterns, while the eruptions producing the Y-2, Y-4, and V-1 (representing the deep-sea equivalent of the lower Thera ash, Bu; McCoy, in press) ash were apparently similar in volume of ejecta and intensity. Most eruptions appear to have been considerably smaller, on the other hand, although distribution patterns for some of the resulting ash deposits are indefinite due to inadequate sampling in cores. In the scale of human values, however, the ca. 1450 B.C. eruption that deposited the upper Thera (Z-2) ash must have been quite significant and devastating then, as it would be now.

CONCLUSIONS

Evidence from deep-sea sediments and from land indicates that the upper Thera (Minoan; Z-2) ash had an easterly to east-south-easterly dispersal pattern due to transport by stratospheric winds. Significant modification of this pattern resulted from atmospheric winds, oceanic currents, and post-depositional processes, the latter being especially noticeable from present-day variations in ash layer thickness. The eruption was not a particularly large one in geological terms in the Mediterranean area, by comparison with the distribution patterns for other tephra layers deposited during the past 240,000 years.

No evidence exists for multiple air-fall episodes during the Minoan eruption. Data are sparse, but it appears that those few cores south of Santorini in the Aegean Sea that appear to have more than one distinct ash layer within the stratigraphic interval where the upper Thera ash is present, may reflect post-depositional mixing and modification of a thicker tephra layer. In comparison with typical eruptive sequences of Plinean-type eruptions, only one ash-fall phase would be expected.

In human terms, however, the eruption was significant. The easterly dispersal pattern implies a significant ash fall of 10 to 20 cm (20 - 40 cm, corrected) over southwestern coastal Anatolia, Rhodes, and other islands to the east of Santorini. Ash fall over Crete produced only about 0.5 to 2 cm of ash, or 1 - 4 cm on a corrected basis, neither thickness of particularly devastating proportions to inhabitants there, at best only a nuisance. While the effects due to earthquakes and tsunami related to the eruption are difficult to assess, it would appear that they could have been much more significant in causing physical damage on Crete. Future efforts towards identifying the upper Thera ash on land and determining cultural consequences attributable to the ash fall should be directed to those islands and continental areas east of Santorini.

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

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