The Application of Tree-Ring and Ice-Core Studies to the Dating of the Minoan Eruption

However, the underlying assumptions of these studies (that there is a causal relationship with the Minoan eruption, and that this is a unique solution) have still to be corroborated. Two lines of argument suggest that these premises are flimsy:

1. The (realistic) range of calibrated ¹⁴C dates currently gives a scatter in the timing of the eruption of ~75-100 years. On the basis of current eruption rates there could have been 20-30 northern hemispheric eruptions of sufficient 'size' to cause any of the observed effects during this period.
2. These studies also assume that, since the Minoan was one of the volumetrically largest (documented) eruptions of recent millennia, it should be assigned to the most dramatic acidity peak, or smallest tree-ring growth rate during the period of study (~1300-1800 BC). However, eruption volumes are only a poor indication of their potential environmental impact. There is increasing evidence that when the eruption did take place the sulphur yield was only of the same order (or less) as ~5 eruptions during the past century. There are plausible alternative candidates for the effects of an eruption of this impact throughout the 16th and 17th centuries BC in both the ice-core and tree-ring records.

INTRODUCTION

Considerable interest has been focused on the timing and effects of the Bronze Age eruption of Thera. For numerous reasons, a knowledge of the precise timing of the eruption will be of great value to interested workers. Much archaeological debate has been directed towards an assessment of the implications of high-precision recalibration of the scattered ¹⁴C dates for the extant chronology established on the basis of Aegean pottery styles. At the same time the results from more indirect, and yet potentially more precise, dating tools have been used to further the arguments in favour of a 'high' chronology. This may have been premature, since the relationship between the observations on which these new data are based and the eruption of Thera remain obscure.

This contribution will not consider the dating of the eruption by ¹⁴C methods since these are comprehensively reviewed elsewhere (for example Manning 1988, 1989 and references therein). It will be restricted instead to a consideration of two potentially valuable techniques (ice-core and tree-ring dating) which are subject to a major assumption: that the Minoan eruption had a massive global impact, the most important between ~1300 and 1800 BC. While it is accepted that the observed effects could have a link with an eruption (or at least a northern hemisphere climatic 'event' in the case of the tree rings) in general (Pyle 1989), it is not clear that it was the Minoan eruption in particular. This relationship must be established before the precise dates possible with these techniques can be used as a method to which importance comparable to more traditional archaeometric methods can be attached.

With the present poor resolution and lack of consensus on how to treat the ¹⁴C dates obtained up till now, it is also not yet sufficient to argue that the several events (ice-core acid peak, frost damage in the western USA and Ireland, and the calibrated ¹⁴C date for the eruption) confirms a causal relationship.

EXPLOSIVE VOLCANIC ERUPTIONS AND THEIR IMPACT

Volcanoes are an important source of a number of atmospheric pollutants, principally silicate ash and acid aerosols. The most important environmental effect of large-scale volcanic eruptions derives from their contribution to the stratospheric aerosol; in particular to the inventory of sulphates. Theoretical studies predict a strong correlation between the stratospheric acid aerosol mass and the concomitant surface temperature drop (Pollack et al. 1976).

While ash particles have little influence, falling out in days to weeks, the sulphur oxidizes and hydrates to a sulphuric acid aerosol (few tenths of micron size) which may linger for months.

Absorption and scattering of incident solar energy warms the stratosphere, but leads to a net cooling effect at the surface of the earth. The cooling influences will last for the same order (< 1-3 years) as the stratospheric residence time of the aerosol. This is supported by observations and reconstructions of major eruptions (e.g. Self et al. 1981; Stothers 1984) and comparisons of estimated sulphur output and recorded temperature changes for various eruptions (Devine et al. 1984).

MEASURES OF ERUPTION SIZE AND ERUPTION FREQUENCY

For the purposes of this discussion it is useful to have an idea of the concept of eruption 'size'. Several indices of and eruption frequency rank have been used to describe the scale of volcanic eruptions, including 'dispersive power' (area affected by fallout), 'destructive potential'  (human, agricultural impact) and 'violence' (Walker 1980). The most valuable measures are the magnitude, or total eruptive mass, and the intensity, the peak effusion rate during the plinian (vertical eruption column) phase.

While 'magnitude' cannot generally be measured without a great deal of uncertainty due to the vagaries of erosion, deposition in the sea and inaccessibility, 'intensity' can be modelled satisfactorily using the on-land dispersal of dense fragments deposited from the eruptive plume (Carey and Sparks 1986). The intensity measured in this way has a simple (predictable) relationship to the eruptive column height and correlates very well with the eruption magnitude, at least for major eruptions (Carey and Sigurdsson 1989); so for ancient volcanic deposits there is a way of determining quantitatively some critical parameters relating to eruption style.

A quantitative index of intensity (similar to the Richter scale for earthquakes) has yet to be widely applied because of the paucity of reliable data (Fedotov 1985). However, there already exists a simple (semi-quantitative) index of magnitude: the 'Volcanic Explosivity Index' (VEI) of Newhall and Self (1982). This, though imperfect, facilitates comparison of eruptions and broadly relates to eruption column heights: any eruption generating a plume which reaches the stratosphere (above ~7-10 km at the poles, ~14-16 km at the equator) is defined to have VEI ≥ 4. It is these eruptions that will contribute to the stratospheric sulphuric acid inventory, have long aerosol residence times (10-12 months, cf. Fuego 1974, El Chichon 1982; Sedlacek et al. 1983; Hofmann 1987) and deposit acid across wide areas. Under suitable conditions this fallout may be preserved in ice cores, or lead to transient (1-2 years) cooling events.

The VEI has the further advantage that it is used in the most recent index of volcanic eruptions (Simkin et al. 1981). This catalogue reports all known eruptions since 8000 BC. Its comprehensive coverage allows us to make estimates of prehistoric eruption rates, from analysis of the well-known recent eruption rates. Since AD 800 (a period over which the majority of eruptions of VEI ≥ 4 will have been recorded) eruptions producing stratospheric columns have occurred at a rate of 3 per decade. Assuming a similar rate for the likely maximum ¹⁴C range of 1675-1525 BC, the Minoan eruption would represent only 1 in 45 northern hemisphere eruptions with VEI ≥ 4, and one of five with VEI of five or larger. An uncertainty of 75-100 years reduces this to 1 of 20-30 eruptions (VEI ≥ 4), and 1 in 3 (VEI ≥ 5).

However, not all of these will have the same potential climatic impact, since large eruptions are not necessarily of sulphur-rich magma, while the chances of preservation (hence, later recognition) as acid fallout is further influenced by the latitude of the eruption. For example, Legrand and Delmas (1988) found that only a quarter of historic eruptions (VEI ≥ 4) in the southern hemisphere left major acidity peaks in an Antarctic ice core, while Hammer (1984) reports a systematic bias towards the products of Icelandic eruptions in Greenland cores. This reflects the influence of pre-eruptive sulphur content and proximity rather than eruption magnitude.

THE VOLCANIC CONTRIBUTION OF SULFUR TO THE ATMOSPHERE

Having established the climatic importance of sulphur, it is instructive to compare estimates of volcanic output with those contributions from other sources. Table 1a summarizes recent estimates of the global atmospheric inventory of sulphur, normalized to the equivalent mass of H₂SO₄. It is clear that the estimated output of actively degassing volcanoes is small compared to anthropogenic and biological productivity, with volcanoes contributing only 4-7 % of the total. Around 50 % of the volcanic output is estimated to be contributed frequent by inmajor eruptions (Stoiber et al. 1987). Most of the atmospheric inventory reaches only the lower troposphere, where it has a very short residence time (hours to days) and little net effect. The infrequent volcanic contribution is of most importance at higher levels in the stratosphere, where it is estimated that > 60% of the input of sulphur can be attributed to major volcanic eruptions (Sedlacek et al. 1983). 

The estimated output for several of the largest eruptions of the past two centuries is shown in Table 1b. The estimates for the Tambora (1815) eruption indicate that this single input probably contributed by up to 50% of the total inventory of atmospheric sulphur for that year; Agung (1963) and Krakatau (1883) were an order of magnitude smaller. From these and other data it appears that an input of ≥ 1 megatonne of sulphates into the stratosphere will have measurable climatic effects.

There is very little experimental evidence by which we may simply calculate the likely loss of sulphur from magmas during eruption (other than by direct measurement), save for the work of Haughton et al. (1974). They found that for silica-poor melts, the strongest control on dissolved sulphur content was the presence of ferrous iron. Extrapolating this to high-silica magmas (such as the Minoan magma), which are poor in iron, would suggest that they should also be relatively poor in sulphur even before degassing. This appears to be the case for the Minoan eruption from the measurements of Devine et al. (1984) and Sigurdsson et al. (1990): the difference between the undegassed glass inclusions and the erupted magma is only ~20 ppm, compared to a difference of 700-900 ppm for the 1783 Laki basaltic magma (Sigurdsson 1982) or 2700 ppm for the Fuego 1974 andesites (Rose et al. 1982). Thus, since both the magnitude and sulphur loss (per unit mass) of 'big' historic eruptions can vary by two orders of magnitude (i.e. 0.3-30 km³ magma, 2000-20 ppm S) and these two are generally inversely correlated, we may expect moderate volume S-rich eruptions of basaltic to andesitic centres to be most readily recorded in proxy climatic records (cf. Rampino and Self 1984).

Using the most recent estimates for the mass of silicate melt alone erupted during the Minoan eruption of 53 X 10¹² kg (Pyle 1990), and the known loss of sulphur (from above) gives a sulphuric acid yield for the whole eruption of ~3 megatonnes. This figure has a likely uncertainty of a factor of two, but is nevertheless substantially smaller than that required by Hammer et al. (1980) to create a substantial acid spike in Greenland. Table 1c summarizes the various estimates of (equivalent) sulphuric acid output for the eruption.

ESTIMATING GLOBAL VOLCANIC ACID OUTPUT FROM ICE CORES

In the past, Hammer and co-workers (1980, 1985, 1987) have used the distribution and size of sulphuric acid peaks in ice cores to estimate the global fallout from volcanic eruptions. This they did by calibrating the fallout at any point against the deposition of the nuclear-test product ⁹⁰Sr during the early 1960s. Simple correction factors for the three major ice cores in Greenland have been established in this way (see Table 2).

This method can be inverted: given an estimate of the likely magnitude of the sulphur output of an eruption, we can calculate the size of the related acidity peak in the Arctic. These calculations are shown in Table 2, for the likely 'worst case' with 8-10 X 10⁹ kg of H₂SO₄ deposition following the eruption. Given an output of this magnitude, it is apparent that the largest peak expected would be only 0.5 μequivH+/kg ice above a typical background of 1. This is a factor of 8-20 smaller than either of the proposed correlative peaks. Suspicion must fall on higher latitude, smaller magnitude, more sulphur-rich eruptions (for example in Iceland) as the sources of these large peaks. A similar argument has been used to relate the acidity levels in Greenland dating from 1963/1964 to the ongoing eruption of Surtsey (Iceland) rather than the proposed correlation of Agung 1963 (Rampino et al. 1988).

Returning to the core acidity profile (Hammer et al. 1987, Fig. 2), there are a large number of peaks of the required order of magnitude between 1525/1675 BC, any of which could equally plausibly represent Minoan fallout.

POSSIBLE CLIMATIC RAMIFICATIONS OF THE MINOAN ERUPTION

Estimates of the total acid output from the Minoan eruption indicate that this was probably of the same order of magnitude as the Agung (1963) and Krakatau (1883) eruptions. Published studies on the effects of these historic eruptions (reviewed in Rampino et al. 1988) give some indication of the likely after-effects of the Minoan eruption. Both were followed by a short period (1-2 years) of cooling, with temperature drops of up to 0.3° C (in accord with theoretical predictions and measured optical depths). Both also correlate with 'frostring' markers in Californian and Nevadan bristle-cone pines (LaMarche and Hirschboeck 1984). These temperature shifts were a factor of 2-3 smaller than those recorded following the particularly sulphur-rich eruptions of Laki (1783) and Tambora (1815), and were still well within the natural scale of variation, which may be equally strongly modulated by other non-volcanic influences, such as the El Niňo oscillation (Angell and Korshover 1984).

On this basis the same scale of effects should be expected for the aftermath of the Minoan eruption. However, it does not follow that these effects should be the largest observed between the 14th and 18th centuries BC: Baillie and Munro (1988) described retarded growth in bog oaks that not only appeared to extend over a period of up to a decade, but which were also the most severe in the lifetime of each tree examined. No study has yet documented a timespan for inclement weather of volcanic origin of this magnitude. The caution expressed by LaMarche and Hirschboeck (1984) when tentatively correlating the Minoan eruption should not be ignored.

TOWARDS A RESOLUTION?

There are several conceivable ways of resolving the true nature of the postulated relationship between the various observed effects and the Minoan eruption. Firstly, there are the new ¹⁴C data expected from the AMS studies, which may enhance the resolution of the dating of the eruption and provide a tighter calendrical control. As the age range for the eruption is reduced, arguments based on the likely rate of eruptions will diminish correspondingly; it will also become clearer whether or not the apparent close temporal relationships are real. Nevertheless the potential rate of three substantial eruptions per decade still stands.

The calculations on the likely impact of the eruption, and the small acidity peak expected in Greenland ice cores will still stand, even if the absolute dating of the eruption is tightened. It is unlikely that new finds of distal ash or improved measurements of volatile loss will shift the calculations here by more than a factor of two.

Analysis of the particulate material (if any) associated with acid peaks over the time range 1500-1700 BC would help constrain the likely sources of the sulphuric acid; this has been done with positive identification of El Chichon fallout in Greenland and Tambora ash associated with the suspected acid peak in Antarctica (de Angelis et al. 1985). This would at least distinguish between basic (low SiO₂) or silicic eruptions and possibly confirm the suspicion that Icelandic signatures dominate Greenland fallout.

Isotopic studies on, for example, the composition of the sulphur are also unlikely to be of much assistance: there is only a small corpus of δ³⁴S data from young volcanoes, with no clear method (as yet) of distinguishing eruptions from different locations. Other radiogenic isotopes (for example Sr, Nd) are also likely to be similarly uninformative, with the problem in these cases that the dominant signature would come from continental dust particles rather than the minor volcanic ash component. Such analysis will only become feasible with the further development of micro-analytical procedures.

CONCLUSIONS

There are a number of substantial criticisms that can be levelled at the proposed correlations between acidity peaks in ice cores, retarded growth recognized in tree-ring sequences and the Minoan eruption of Santorini:

1. Eruption rates are such that, with an uncertainty in the age of the eruption of 100 years, there would have been 30 eruptions during that time with the potential to reach the stratosphere and cause the observed effects.
2. The likely acid yield from the eruption would in any case have left a peak of ≥ 0.5 μequivH+/kg ice, or an order of magnitude lower than either of the proposed correlative peaks.
3. There are alternative candidates, of a smaller magnitude than proposed, in both the ice cores and tree-ring sequences, which could equally well relate back to the eruption of Thera.
4. In the absence of other corroborative evidence, there seems no strong reason at this time for attributing a date of 1645 BC or 1626-1628 BC to the Minoan eruption on the basis of either tree-ring or ice-core chronology alone.

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

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