The Precision of Ice-Core Dating

A strong volcanic acid signal in an ice core can only be related to a specific eruption if other data or information on the eruption exist. This kind of additional information must be of such a nature that it verifies the ice-core information. Such information exists in the case of the 1645 BC signal, but the authors placed a question mark in the title of the paper, because they could not exclude the possibility, that the volcanic signal was caused by another - as yet unidentified - eruption.

Criticism has been levelled at the interpretation of the Dye 3 volcanic signal, but to the authors' surprise, the main criticism concentrated on the precision of ice-core dating (e.g. Hughes 1988): the main problem being that the frost-ring events in trees gave another date for the eruption.

In this paper we concentrate on the dating precision of polar ice cores back to around 2000 BC, even though the principles are valid and equally relevant for dating of polar ice cores some 10,000 years back in time. Our aims can be summarized as follows:

1. To present the advantages and problems of polar ice-core dating, in order to communicate to non-glaciologists a better understanding of the dating procedures.
2. To explain why we are reluctant immediately to accept the tree-ring date for the Thera eruption.

Further, we will discuss the dating of the Thera eruption in connection with the new planned deep drilling in central Greenland, i.e. the Greenland Ice-core Project (GRIP), which has been funded for the period 1989-1992.

INTRODUCTION

Ice cores from the central parts of the remote and highly elevated polar ice sheets have proven to be valuable sources of information on palaeoclimatic and palaeoenvironmental conditions.

The interpretation of the data obtained by ice-core analysis relies on some general ideas and/or models, which relate the ice-core data to the composition of the atmosphere (over the ice sheet) and subsequently to the sources of atmospheric trace substances.

The interpretation is often complicated and depends on the investigated parameters (isotopic composition of the snow, soluble and insoluble matter). The interpretation of environmental records in glaciers and ice sheets has recently been the subject of a Dahlem workshop (Oeschger and Langway 1989). Below we will concentrate on the ice-core dating of the Thera eruption (Hammer et al 1987).

ICE-CORE RECORDS AND PAST VOLCANISM

The polar ice-core record of past volcanism is a special example of the environmental record in glaciers and ice sheets and it is possible to relate the acid ice-core signals to major, historically known eruptions (Hammer 1977; Hammer et al. 1980). In some cases it has been possible to estimate, within a factor of 2, the total volcanic acid gas production (H₂SO₄, HCl and HF) by comparing the spread and transport of the acid volcanic gases to the similar process of spread and transport of fission products (Fig. 1) from the nuclear bomb tests in the 1950s and 1960s (Clausen and Hammer 1988). Mega-eruptions (Volcanic Explosivity Index VEI > 6 are often untypical and therefore give rise to larger uncertainties in estimates of fall-out patterns. This applies to the spread and distribution of gaseous products from the Minoan eruption (VEI 6.9), and may also apply to the uncertainty in estimates of the amount of emanation products (e.g. sulphur dioxide) deduced from petrological analyses.

The ice record from the Dye 3 core shows a number of major volcanic signals, which were caused by yet unidentified volcanoes. The eruptive ice-core signal in 1645-44 BC, which has been suggested to be caused by the Thera eruption (Hammer et al. 1987), would have belonged to this group of eruptions, had it not been for the special character of the ice record over the years 1900-1300 BC where only a single major volcanic event can be traced, as described by Hammer et al. 1987.

With the present available techniques for ice-core analysis the authors could not prove, but only suggest the eruptive signal to be caused by the Thera eruption; hence they placed a question mark in the title of their paper.

That the acid signal in 1645-44 BC was caused by a major volcanic eruption is beyond doubt; more moderate volcanic eruptions could hardly mark the annual precipitation through nearly two years unless extreme amounts of acid gases were produced and injected into the stratosphere. The time variation of the acid signal also indicates a northern mid-latitude eruption. Nelson et al. (1990) argue, that there may have been several other major eruptions which could have caused the ice-core signal in 1645-44 BC and the frost event in tree rings in 1627 BC (LaMarche and Hirschboeck 1984). If so, it must be concluded that out of several major eruptions which took place during 1900-1300 BC only one produced sufficient amounts of acid gases to clearly mark the precipitation falling on the Greenland ice sheet. Over the past 1000 years volcanic eruptions with Volcanic Explosivity Index VEI > 5 (Simkin et al. 1981) have left a clear signal in polar ice cores. This is also supported from a geological point of view (Pichler and Friedrich 1980). Hence, if the signal in 1645-44 BC was not caused by the Thera eruption, it is of interest to obtain documented volcanological evidence for the existence of other eruptions with VEI > 6 over the time period in question.

ICE-CORE DATING

An important prerequisite for stratigraphical ice-core dating is that the ice core originates from a region which receives sufficient precipitation, i.e. the annual snow accumulation is distributed evenly over the year at the drill site. This is not always the case, but with respect to the dating of the 1645-44 BC volcanic event we are not concerned because the rate of annual precipitation at Dye 3 is quite high (55 cm of ice equivalent per year) and the south Greenland ice sheet receives some 20-30 major snowfalls distributed over the year.

The stratigraphical dating precision also depends upon the number of ice-core parameters measured; this may vary from core to core, but in the case of the Dye 3 core, three independent techniques were applied, i.e. stable isotopic composition of the ice, insoluble dust concentration and acidity along the core (Hammer et al. 1986).

To some extent stratigraphical ice-core dating is based on the experience obtained by the dating of the last 1000 years, which are the best documented years. For these years a multitude of cores have been recovered and they have generally been dated by seasonal stratigraphy without the use of reference horizons. Dating of older annual layers follows the same principles, because the seasonal character of the dating parameters have only changed little over the past 10,000 years.

If a high data resolution is obtained along the ice core, as may be the case with dust and acidity measurements, the record approaches a time resolution which is so detailed that it reflects individual precipitations within a single year. Such records, when compared, can lead to a very high dating precision. Fig. 2 and 3 show examples of highly resolved dating data.

The seasonal variation of e.g. δ¹⁸O, dust and acidity are to a large degree independent. Each of these three parameters could serve as a stratigraphical dating parameter, but combined they offer a very strong and precise dating method.

The dating of the Dye 3 deep core was complicated by melt layers and a variable local/regional, year-to-year, influence of the meteorological conditions which are typical for the south Greenland ice sheet. Hence the dating of the Dye 3 core, even though rather straightforward, was not an ideal case and the dating of the 1645-44 BC event - the supposed Minoan eruption - called for some caution with respect to the dating precision. A statistical treatment of the data was undertaken, but it was decided to present the dating in a way which encompassed all reasonable doubts and uncertainties of the dating precision. This is what is meant by the error limit. The statistics may be improved as described by Manning (1990) but it is more important to obtain better material for C-14 dating as demonstrated by Friedrich et al. (1990).

ICE-CORE AND TREE-RING DATES

The frost events in tree rings have suggested a date for the Minoan eruptions of 1627 BC (LaMarche and Hirschboeck 1984) and it is tempting to compare this event to the ice-core volcanic signal in 1645-44 BC, as suggested by Hughes 1988. However, we are reluctant to accept that the two events occurred at the same time. The error limit quoted for the ice-core date is certainly not the most probable error, even though the error might be as high as ± 20 years. Further, LaMarche and Hirschboeck observed frost damage in tree rings, which could not be related to major volcanic eruptions; also some major volcanic eruptions, e.g. the one in AD 1259, comparable to the AD 1815 Tambora eruption, did not give rise to any frost damage in tree rings. The latter eruption marked the Antarctic and Greenland ice sheets with highly elevated acid ice layers as exemplified in Fig. 4 (Langway et al. 1988).

In our opinion the ice-core records on past volcanism reflect the actual global or hemispheric volcanic dust veil better than does the less objective volcanic record used by LaMarche and Hirschboeck. The more recent data on Irish tree rings (Baillie and Munro 1988) do not change this conclusion.

For a final settling of the question we prefer to await the dating of the new deep ice core planned to be drilled during 1990-92: this deep drilling project, GRIP, has already started in 1989, but the actual drilling will first take place in 1990 and the following years.

If two well-measured deep cores exist, i.e. the Dye 3 and the new deep core, it will be possible to improve the dating precision, because dating problems in one of the cores can to a large extent be solved by comparing the two cores year by year. Based on our experience from more shallower cores we estimate that with two cores available, an error limit of ± 5 years may be obtained for the 1645 BC event.

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

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