Previously on a University Of Wales, Aberyswyth Web Server

Lecture two: Climatic responses to volcanic eruptions.

Summary of Lecture.
This chapter discusses the theoretical and empirical framework on which much of the debate which associates volcanic eruptions and climate change has been based. Historical records are used to suggest that there is some justification for the assumption that a volcanic eruption may bring about climate change, but the often loose (and convenient) correlation of historical events and volcanic eruptions is challenged. In this chapter The use and construction of various eruption chronologies are considered and it is demonstrated that many of them are inadequate. Alternative records of volcanic activity, in particular ice-core acidity are considered. Measures of a volcanic eruption's ability to effect climate change are discussed. In particular the "Dust Veil Index" [DVI] and "Volcanic Explosivity Index" [VEI] are examined. As an estimate of the potential of an eruption to perturb climate, the DVI is shown to be inadequate in that it over emphasises the importance of dust output and employs a circular argument to establish the relationship between "climatic flux" and eruption magnitude. The degree to which temperatures have responded to Twentieth Century A.D. eruptions is considered and the mechanisms by which a volcanic eruption may perturb climate are discussed.

Why have volcanic eruptions been identified as having the potential to effect environmental change? We need to understand the climatic mechanisms by which environmental change may be caused in regions which are distant from the eruption. In particular can specific eruptions trigger the global and climatic phenomena described by Baillie [1988, 1989 a & b; 1991 a & b] and Pang et al. [1987, 1988, 1989]?

Volcanic eruptions and surface cooling.
The assumption that cooling will follow a volcanic eruption is based on observations made by Benjamin Franklin in Paris, in 1783. Franklin described atmospheric and climatic phenomena, which we now know were the result of the Laki fissure eruption in Iceland.

"During several of the summer months of the year 1783, when the effect of the sun's rays to heat the earth in these northern regions should have been greatest, there existed a constant fog over all Europe and great part of North America. This fog was of a permanent nature; it was dry, and the rays of the sun seemed to have little effect towards dissipating it, as they easily do a moist fog arising from water. They were indeed rendered so faint in passing through it that when collected in the focus of a burning glass, they would scarce kindle brown paper. Of course, their summer effect in heating the earth was exceedingly diminished. Hence the surface was early frozen.
Hence the first snows remained on it unmelted, and received continual additions. Hence the air was more chilled, and the winds more severely cold. Hence perhaps the winter of 1783-4 was more severe, than any that had happened for many years.

The cause of this universal fog is not yet ascertained. Whether it was adventitious to this earth...or whether it was the vast quantity of smoke, long continuing to issue during the summer from Hecla in Iceland, and that other volcano which arose from the sea near that island, which smoke might be spread by various winds over the northern part of the world, is yet uncertain. It seems however worth the enquiry, whether other hard winters, recorded in history, were preceded by similar permanent and extended summer fogs."
[Franklin 1784].

Franklin's observations make it possible to suspect and even detect the impact of other volcanic eruptions in extant historical records. In 44 B.C., Plutarch described the veiling of the sun's rays

"..the obscuration of the sun's rays. For during all that year its orb rose pale and without radiance, while the heat that came down from it was slight and ineffectual, so that the air in its circulation was dark and heavy owing to the feebleness of the warmth that penetrated it, and the fruits imperfect and half ripe, withered away and shrivelled up on account of the coldness of the atmosphere"
[Plutarch cited in Stothers & Rampino 1983a].

Not all volcanic eruptions are so clearly identified with atmospheric veiling, but a strong ice-core acidity signal [Herron 1982] suggests that an unknown volcanic eruption was responsible for similar phenomena in the year A.D. 536:

Similar accounts can be found in Chinese historical annals [Pang et al. 1987, 1988, 1989]. One such has been dated to 1600 ± 30 BC, and correlated with narrow tree-rings in the Irish Record [Baillie & Munro 1988], frost-damaged tree- rings in California [LaMarche & Hirschboek 1984], and an acid peak in the Greenland ice-core [Hammer et al. 1987].

"At the Time of King Chieh the sun was dimmed ... Winter and summer came irregularly ... Frosts in the sixth month ... The five cereals withered ... therefore famine occurred" [Pang et al. 1989]

The common themes in the accounts above are a reduction in the energy of the sun's rays as they reach the earth and a veiling or dimming of the sun. The comments of Franklin in 1784 allow us to suggest that these phenomena indicate the presence, in the atmosphere of volcanic aerosols. It would be easy to assume that similarly cool summers and harsh winters generated the environmental stress indicated by the Irish tree-ring record. A severe, if unspecified, environmental deterioration could then be postulated with some degree of "theoretical confidence". But is the dating framework secure enough to allow this interpretation to be suggested with confidence? Are such descriptions firm enough to allow us to construct models of critical environmental change in specific geographical areas, in which volcanic eruptions appear to be the only independent variable?

Correlating historic climate records and volcanic eruptions: the need for caution. The need for caution can be demonstrated by further investigating one of the cases cited above, for which the Chinese "Annals", Irish and American dendrochronological research and ice-core evidence appear to be in agreement in recording the impact of a major volcanic eruption, the eruption of Thera. It was suggested that several lines of evidence, tree-ring width minima and ice-core acidity peaks, have actually recorded the impact of the eruption of the Greek island of Thera, in Minoan times [Hammer et al. 1987; Manning 1988]. Pang et al's. [1989] estimated date for this event was reached by "counting back" the generations from a "secure date" of 841 BC. To the cynical this gives a sixty year window around 1600 BC within which any convenient volcanic eruption may be "blamed" for the "impacts" detected. It is unrealistic to suggest that only one significant volcanic eruption, that of Thera, occurred within this time frame [McClelland et al. 1989]. Vogel et al. [1990] and Beget et al. [1992] list at least three other major eruptions which took place within this period: Avellino in Italy, Mt. St. Helens in Washington State, and Aniakchak in Alaska. To further disturb the "Theran" argument, the dating of the acid peak in the Greenland ice-core record has been revised from 1390 ± 50 BC [Hammer et al. 1980] to 1645 ± 7 BC [Hammer et al. 1987]. Subsequently Hammer and others [1988], Cadogan [1988] and Pyle [1989], suggested that it was unlikely that the eruption of Thera was the culprit for all or any of the phenomena which researchers were attempting to assign to it. A strict interpretation of dates places the ice-core acidity peak outside the time frame suggested by the Chinese Annals. The "absolutely" dated tree-ring records indicate environmental stress in the period 1628-1626 B.C. [LaMarche & Hirschboek 1984; Baillie & Munro 1988], which is at least 10 years after the ice-core acidity peak date, and only just within the sixty year range suggested for the poor summer in China. The desire to link such phenomena and the stretching of the dating frameworks involved is an attractive but questionable practise and one which is discussed in depth by Baillie [1991 a & b].

All such attempts to link (and hence infer associations between) historic eruptions and environmental phenomena and human "impacts", rely on the accurate and precise association in time of the two events. The account above suggests that this does not occur for the eruption of Thera. A more general investigation of eruption chronologies constructed since 1970 suggest that such associations are frequently unreliable when based on eruption data gathered earlier than the twentieth Century.

Eruption Chronologies.
The use of eruption chronologies to prove that volcanic eruptions are the major independent variable in the generation of transient but powerful "fluctuations" such as cooler temperatures, damaged tree-rings, disturbed atmospheric circulation patterns, famine and social unrest, remains problematic. The mere association of an eruption and an environmental phenomenon thought worth recording in the past is not proof of a dependent relationship. "There is no question but that the available data suggest an unmistakable coincidence in time between the occurrence of major volcanic eruptions and an accompanying or subsequent minimum in curves of surface temperature. The important question remains whether this represents a causal relationship or a mere coincidence in time". [Ellsaessar 1986 p.1184].

There is a danger that any historical or environmental change may be ascribed to a convenient volcanic eruption. As Renfrew [1979 p.582] notes: "... it is necessary to recognise and discount the common tendency among archaeologists and historians to assume a causal link between the distant and widely separated events of which they may have knowledge. An eruption here, a destruction there, a plague somewhere else - all are to easily linked in hasty surmise"

A review of four eruption chronologies constructed since 1970 will illustrate the problem [Figures 2.1 & 2.2]. In 1970, Lamb's seminal work on volcanic activity and climate change published an eruption chronology for the years 1500 - 1969. This recorded 380 known eruptions. Ten years later Hirschboek [1980] presented a revised eruption chronology which recorded 4796 eruptions for the same period, a very significant increase on Lamb's figure. Simkin and others [1981] raised this figure to 7664 eruptions and Newhall and Self [1982] further increased the number of known eruptions in this period to 7713. While the upgrading of eruption chronologies presents a manageable problem the deterioration in the temporal quality of the data is less easily accounted for. The difference between the number of eruptions recorded in the twentieth century and in the sixteenth century can be seen on Figure.1 and 2 and in Table 1. The marked rise in the number of volcanic eruptions recorded in each century is a factor of better reporting rather than a real increase in eruptive activity. The paucity of eruption data in earlier periods is illustrated in Figure 2.1 which shows that 3018 eruptions were recorded between 1900-1969 and 11 were recorded between 1 and 100 AD. [Simkin et al. 1981]. The data is also weighted, for obvious reasons, by observations made in Europe [Stothers & Rampino 1983 a & b] and in China [Pang et al. 1986, 1987, 1989].

The role and application of the Volcanic Explosivity Index [VEI] will be reviewed below but recent work on eruption frequency and scale in the Twentieth century A.D. [McClelland et al. 1989], indicate eruption frequencies per 1000 years of 12 000-VEI 1, 12 000-16 000-VEI 2, 4000-5000-VEI 3, 600-VEI 4, 65-VEI 5, 25-VEI 6 and 1-5-VEI 7. This gives a total expected eruption rate of approximately 30 000 per 1000 years. The fact that only 19 eruptions > VEI 5 are known of [Simkin et al. 1981], out of an expected frequency of 100, shows that caution must be used in the application and correlation of eruption chronologies and other "proxy" data.

The poor nature of the data makes any attempt to prove an association between recorded historical eruptions and weather by statistical analysis a doubtful exercise. Equally problematic is any attempt to suggest that mere association is sufficient to suggest that one of the few recorded eruptions might be responsible for archaeological, palaeoenvironmental or historical phenomena.

Within the twentieth century A.D., despite improved reporting and recording of eruptions, volcanic events have sometimes gone unrecorded. Simkin et al. [1981 p.25 their Figure 5] show that the number of recorded eruptions dropped steeply during the First and Second World Wars. Hoyt [1978] suggested that an unrecorded major eruption in 1928, of Paluweh on the Lesser Sunda Islands, Indonesia, was responsible for the reported atmospheric turbidity and a decrease in Northern Hemisphere summer temperatures of 0.3°C. Sedlaek and others [1983] monitored sulphur concentrations in the stratosphere between 1971-81 and noted several injections of volcanic sulphates into the stratosphere for which no eruption had been recorded or reported in the literature.

This analysis shows that the dating framework within which volcanic events and historical or environmental records of climate change are related cannot be regarded as accurate. It is not sufficient to assume a causal relationship between an eruption and an environmental response. Not all eruptions modify climate and it is not sufficient to model an unspecified, but large, volcanic event, and an unspecified, but severe, degree of climate change. To deal with this problem an eruption must be categorised, the size of an eruption must be estimated, and the degree of climatic change expected to follow modelled. Two attempts to do this are the work of Lamb [1970] who constructed a measure known as the DUST VEIL INDEX, and Newhall and Self [1982] who constructed the VOLCANIC EXPLOSIVITY INDEX.

An attempt to quantify the "climatic effectiveness" of volcanic eruptions was made by Lamb [1970, 1972, 1977], who worked on the principle that the passage of the sun's rays through the atmosphere was primarily inhibited by the volume of dust emitted in an eruption. The climatic effectiveness of any single volcanic eruption could therefore be assessed by an measuring, or estimating, its dust (i.e. tephra) output. Lamb's measure of climatic effectiveness was termed the Dust Veil Index [DVI] and has formed the basis for much subsequent work. Focusing on the volume of dust emitted by different eruptions, and using the 1883 eruption of Krakatau as a standard, Lamb constructed a classified chronology of eruptions. The volume of dust emitted in an eruption was correlated with available temperature records, and Lamb's research suggested a positive correlation between low temperatures and a DVI 100.

Lamb emphasised the importance of the role played by the fine dust thrown into the stratosphere during an eruption. Lamb further suggested that stratospheric circulation would pass dust from low latitude eruptions around the world whilst dust from high latitude eruptions would remain in the hemisphere of origin. Different climatic impacts were therefore modelled for eruptions which occurred at different latitudes. Several other controlling parameters were suggested, in addition to the latitude of the eruption, which ought to have a key role in causing climate change, these were; the residence time of the dust in the atmosphere, the size of the dust cloud, the maximum departure from average temperature and the greatest depletion of incoming solar radiation.

It is worth exploring the relationships of the parameters used by Lamb in his formulae as these constitute the basis for much later work. The DVI could be calculated in three ways:

  1. DVI = 0.97RDmax Emax tmo.
  2. DVI = 52.5tdmax Emax tmo.
  3. DVI = 4.4qEmax tmo.
Lamb considered that those eruptions achieving a DVI of 100 or over were capable of a significant impact on climate.

Problems with the DVI
There are problems with the use of Lamb's indices. Values of R, T, and t, are calculated from observations in mid-latitudes but these are not necessarily where the maximum effect of a low or high latitude eruption would be felt. Calculating the DVI of an eruption based on recorded values of T (temperature lowering) could lead to a circular argument. This is because any temperature variation from the norm is directly related to the size of an eruption, and not to any other external forcing factors such as variation in the output of energy from the sun [Eddy 1977, 1992], or fluctuations in the earth's orbit [Kukla 1979]. Problems also lie in the use of an inadequate Eruption Chronology, and in the assignment of a values for RDmax (greatest depletion in Solar radiation) and Emax (latitudinal dispersal of the dust cloud) from proxy records. Inadequate data exists from which to assign a DVI to eruptions prior to the nineteenth century, or to small eruptions, or where several eruptions were closely spaced. As a result there is a real risk of entering into a circular argument where temperature declines are used to support estimates of the value of RDmax, which value is then held as a measure of the degree of eruption magnitude. It is simply not safe to assume that all temperature departures in eruption years are the result of volcanic activity. Estimates of Emax fail to take into account the possibility of other phenomena such as dust storms being interpreted as being of volcanic origin. Of Lamb's 250 DVI estimates only 10% were based on measured observations of a decrease in solar radiation or fall in temperature, whilst 48% were based on non-quantitative descriptions of eruptions such as may be gleaned from travellers' accounts and eye witness observations [cf. Burton 1875; Nordenskiöld 1876] . Problems also arise in the existence, accuracy and relevance of temperature records. Accurate and precise temperature records do not exist beyond the seventeenth century [Manley 1974] and reliance on other types of observations of climatic severity can be misleading. Baillie [1982 p.248] noted three instances of the "worst winter in living memory" within a five year period in Ireland 1766-71. Temperature series constructed from observations made in a continental interior may not be relevant for conditions on the continental fringe and obviously the opposite must be equally true.

In order to assess the value of the DVI as a tool to be used in reconstructing the climatic impact of prehistoric eruptions, several studies which have linked the DVI to environmental change, will be reviewed below.

Applications of the DVI.
Pursuing Lamb's theme, other workers announced significant negative deviations in surface temperature following the injection of volcanic dust into the stratosphere [Mitchell 1971; Mass & Schneider 1977]. In an early attempt to refine the use of the DVI as a major climatic forcing mechanism, Cronin [1971] investigated the periodicity with which volcanoes had erupted in the mid Twentieth Century A.D. He suggested that the eruption of volcanoes at lower latitudes where the stratosphere is lower, had a greater climatic impact than the eruption of volcanoes at high latitudes, where the stratosphere is higher. Cronin's work was undermined by the use of an inadequate "Eruption Chronology". Several significant twentieth century eruptions A.D. were ignored. In particular eruptions of Alaskan volcanoes (Trident in 1952 and Spurr in 1953), and of Bezymianny on the Kamchatka peninsula, in Siberia in 1956, and the Japanese volcano Tokachidake in 1962, were all absent from his analysis. Bray [1974] used the DVI in an attempt to correlate volcanic activity and glacier response in the northern and southern hemispheres over the past 40 millennia. An attempt to prove a monocausal relationship between the two phenomena, this work demonstrates the problems inherent in such an approach. His work was hindered by an inadequate eruption chronology. The response of glaciers to volcanic forcing was inconsistent, in some cases it was synchronous and in others lagged by up to 100 years. Both the accurate dating of volcanic eruptions which occurred up to forty thousand years ago and the estimation of their DVI was problematic. The DVI was inadequate to explain all glacier response and the inconsistent response time suggested further problems with the use of the DVI to account for climate change.

The theme of glacier response to significant DVI events was pursued by Bradley and England [1978]. Between 1947-1963 glaciers in the North American High Arctic experienced a continuous loss of ice at a rate of circa 3500 kg/m2/yr. After 1963, this loss fell to 350 kg/m2/yr. They suggested that the DVI of 800 calculated for the 1963 eruption of the Indonesian volcano Mt. Agung, indicated its responsibility for this phenomenon. As with the work of Cronin [above], Bradley and England ignored or were unaware of significant DVI events between 1947-1963. They also assumed a "temperature response" to volcanic eruptions which was not quantified. In fact, in the first year after the eruption of Mt. Agung, a decrease in northern hemisphere temperatures of no more than 0.42°C is reported and temperatures had returned to normal by the third year after the eruption [Hansen et al. 1978; Newell 1981a; Rampino & Self 1982]. The 1963 eruption of Mt. Agung is therefore unlikely to have been the sole cause of north American glacier fluctuation and the DVI is shown to be an inadequate measure of volcanic ability to force climate change.

There is obviously a flaw in the deterministic application of the DVI in studies of climatic change. Miles and Gildersleeves [1978] suggested that the problem lay in the reliance on questionably complete eruption chronologies, but it may equally lie in the assumption that volcanic eruptions are the sole or major independent variable capable of generating climate change.

The DVI is seen to be an inadequate measure of the climatic effectiveness of a volcanic eruption. In the case of prehistoric eruptions which may have had an impact on the north British environment, the calculation of a DVI would be largely inaccurate. The "responsible" eruptions are not always clearly identified, and to calculate a probable DVI from a climatic response which itself has been estimated as a result of supposedly related settlement abandonment, or tree-ring stress would lead to a circular argument. It is not sufficient to demonstrate that narrow tree-rings apparently correlate with apparently large ash-producing eruptions, which might have caused a climatic deterioration. The DVI is therefore an inadequate measure for such purposes. Other estimates of magnitude were subsequently developed as a result of some of the failings described above.

The Volcanic Explosivity Index.
An alternative measure of volcanic eruptions was proposed by Newhall and Self [1982] who suggested that it was the volume of dust injected into the stratosphere which determined the climatic effectiveness of an eruption and advanced a Volcanic Explosivity Index [VEI], as a measure of this effect. The VEI depends on an estimation of five parameters of magnitude first suggested by Walker [1980, 1981]. These are:


Table 2. The Volcanic Explosivity Index.
As with the DVI, estimates of the VEI are necessarily based on estimates of variable quality, but being based on volcanologically-derived data it removes the circularity of argument which was built into the use of RDmax and Emax in estimating the value of the DVI. Based on these factors a Volcanic Explosivity Index was assigned to known eruptions on a scale of 0 - 8 Table 2. Eruptions assigned a VEI 4 are those which are estimated to have produced at least 108 m3 of ejecta and achieved column heights of 10 - 25 km. These estimates are independent of observations of temperature depression, atmospheric effects, decreases in radiation receipts or by climatic observations in mid-latitudes. As such the estimates of magnitude reached are climatically independent. Attempts to link significant VEI events and climatic change avoid some of the pitfalls which undermined the use of the DVI. Cruz-Renya [1991] subsequently demonstrated that using the VEI, the scale and frequency of volcanic eruptions could be fitted by a Poisson distribution. Thus one eruption of VEI 7 and 100 VEI 4 can be expected in a 500 year period if time is viewed as a statistically "normal" sample population.

While the VEI includes estimates for magnitude and stratospheric penetration, a significant degree of climatic forcing associated with volcanic events assigned a VEI 4 is assumed rather than demonstrated [Hingane et al. 1990; Mukherjee et al. 1987]. In a critical review of the application of this index to studies of climate change, Bradley [1988] found a statistically-significant temperature anomaly when the 44 known events with a VEI 4 between 1883 and 1981 were analysed through a "superposed" epoch analysis. However, the observed temperature reduction was small, between 0.05 to 0.1°C, and seen only in the month following the eruption. When the eruptions were divided into three latitudinal zones, no significant temperature departure in either the latitudinal zone, or the northern hemisphere as a whole was observed. Large, but historically-rare events, such as the 1912 eruption of the Alaskan volcano Katmai, assigned a VEI of 5 or more, were correlated with temperature depressions <0.4°C lasting for 2 to 3 months. Temperature responses of this degree seem an unlikely catalyst for long-term environmental change in northern Britain and there appears to be no evidence that any large eruption, of VEI <4 has had a significant effect on low frequency temperature changes.

Eruption chronologies based on personal observations are inevitably weighted by both time and location, hence the vast increase in recorded eruptions in the twentieth century and the dominance of Mediterranean eruptions and Chinese observations in the early historic records. Poor recording and geographic weighting have hindered attempts to isolate "clear volcanic signals" from the background noise of the environmental and historical records.

Alternative records
Proxy records of volcanic eruptions constructed from material deposited on the Greenland and Antarctic ice cap have may the potential to identify climatically- significant eruptions with greater reliability and precision. The annual deposition of snow in the Earth's polar regions contains a wide range of atmospheric, climatic and environmental information [Lorius 1990]. The accumulation, in Greenland, of annually deposited snow, over a period of 150 000 years, preserves this information in consecutive dateable layers. Though the precision of the record deteriorates with time as it becomes more difficult to identify annual snow deposition, some the acids and particles emitted by some volcanic eruptions will, with luck, be preserved in each "annual" layer. Prominent acidity peaks or micro-particle concentrations, in ice-cores can be used as reference horizons [Langway et al. 1988], and allow an estimate to be made of climatically significant volcanic activity which is not biased in time and space by human "observation".

Acid concentrations within ice cores may be a good source of information on both the occurrence and the intensity of volcanic events. Hammer [1977, 1984] and Hammer and others [1980, 1981] constructed a chronology using acidity concentrations in the Dye 3, Crête and Camp Century ice cores from Greenland. The concentration of these acidity peaks can be used as both an indicator of climatic impact and of global acid fall out. Since it is the volume of acids injected into stratosphere that forces climate change, the acid level in the ice-core offers a means to check the acid output and hence potential climatic impact of historically-famous eruptions, and identify the occurrence of unrecorded events. . Rather than recording all eruptions, ice-core acidity peaks are best viewed as providing a record of the occurrence of volcanic events which have the potential to affect climate [Sigurdsson 1982]. This removes much confusing noise from the Eruption Chronology and records the occurrence of eruptions which may not have appeared in other records. Several such ice-core acidity peaks appear to correlate with ring-width minima in the Irish record [see Chapter One]. For further information Table 3 presents a list of eruptions with significant acid peaks and estimates of their acid fallout in Greenland and on a global scale. Trends in global volcanism can also be detected and correlated with global temperature trends. Ice-core acidity records [Table 3] indicate that the highest rates of volcanic activity since AD 533 occurred in the periods AD 1250-1500 and AD 1550-1700; the initial and culminating phases of the Little Ice Age. Similar work in both Greenland and Antarctica [Dia et al. 1991; Legrand & Delmas 1987] allow the ice core acidity record to be refined and the removal of the inevitable latitudinal bias which the proximity to Greenland of the volcanic areas of Iceland, Alaska and Kamchatka has introduced. Both Dia et al. [1991] and Legrand and Delmas [1987] record a major acid peak in 1810, which preceded the massive eruption of the Indonesian volcano Tambora in 1815 and suggest this pre- Tambora event which is absent from eruption chronologies compiled from "personal observations" and historical records, was on the scale of the 1963 Mt. Agung eruption.

Langway and others [1988] present ice core acidity evidence for an unrecorded acid producing event in 1259 AD, which produced an ice-core acidity signal on the scale of that produced by the Laki fissure an Tambora eruptions. The omission of an event of this scale from eruption chronologies derived from other sources illustrates their failings, and demonstrates the futility of using such compilations to assign statistical significance to the correlation of known eruptions and climatic extremes.

Often the magnitude of acid peaks has allowed an estimate to be made of the comparative magnitude of each eruption. Clausen and Hammer [1988] analysed the acid signal of the 1783 eruption of the Laki fissure and the 1815 eruption of Tambora and suggested that global acid emission for each event was over 200 X 106 tons. The identification of the acid peaks associated with climatically significant events in relatively recent times will allow an estimation to be made of the climatic effectiveness of the smaller eruptions thought to have had a significant impact on the north British environment in the past.

Volcanic acid
fallout in Global acid
Greenland. fallout
kg km2. X106

CRETE. Location. Date
Agung Indonesia 1963 9 20
Hekla Iceland 1947 6 5
Katmai Alaska 1912 37 30
Krakatoa Indonesia 1883 21 55
Tambora Indonesia 1815 58 360
Laki Iceland 1783 116 280
Lanzarote Canary Isl. 1730-36 36 60
Krafla Iceland 1724-30 63 55
Unknown ? 1666?
} 60 100
Pacaya Guatemala 1664
Awu Indonesia 1641
Adiksa Indonesia 1641 } 81 190
Komagatake Japan 1640
Unknown ? 1600/01 61 50
Raudubjallar Iceland 1554 23 20
Unknown ? 1257/58 128 300
Hekla 1 Iceland 1104 51 45
Eldgja Iceland 934 189 165
Unknown ? 622/23 52 45

Camp Century. Location Date BC. Unknown ? 50±30 192 120
Unknown ? 210±30 72 45
Unknown ? 260±30 54 35
Hekla 3 Iceland 1120±50 99 60
Thera Greece 1390±50 98 125
Hekla 4 Iceland 2690±80 96 60
Unknown ? 3150±90 255 160
Mt. Mazama Usa 4400±110 156 200
Hekla 5 Iceland 5470±130 90 60
Unknown ? 6060±140 119 75
Unknown ? 6230±140 102 65
Unknown ? 7090±160 79 50
Unknown ? 7240±160 124 80
Unknown ? 7500±160 51 35
Unknown ? 7640±170 412 260
Unknown ? 7710±170 69 45
Unknown ? 7810±170 73 45
Unknown ? 7910±170 95 60

Table 3 . Magnitude of volcanic events estimated from the annual acidity layers in ice-cores in Greenland. Based on figures retrieved from Hammer et al. [1980] and upgraded after Clausen and Hammer [1988] & Langway et al. [1988]. Micro-particle content.

As with acid output, so the tephra emitted in volcanic eruptions may be deposited in the annual layers of an ice-core. Betzer et al. [1988] discovered the long range transport of mineral particles up to 10 000 km from their source and Ram and Gayley [1991] have discussed the long range transport of volcanic material to the Greenland ice-sheet. Micro particle concentrations in an ice-core may therefore indicate the occurrence of large ash-producing volcanic eruptions.

Several "peaks" in micro particle concentration in an Antarctic ice-core drilled at the Amundsen-Scott South Pole Station [Thompson & Moseley-Thompson 1981] were found to correlate with known volcanic eruptions Krakatoa, Indonesia 1883; Coseguina, Nicaragua in 1835; Tambora, Indonesia in 1815 and Mayon, Luzon in 1766. Examination of particles under a scanning electron microscope [Moseley-Thompson & Thompson 1982] confirmed the volcanic origin of the material. Micro particle concentrations may therefore offer an alternative or complimentary record of eruption history.

Correlating results from Antarctica, Greenland and Devon Island, Danny-Harvey [1988] found that large increases in atmospheric aerosol loading had occurred during the "Late Devesian-Glacial Maximum". The optical depths indicated by the aerosol content of the ice-core were calculated and the results suggested that increases in aerosol content at that time would have lowered the global mean temperature by 2-3 degrees centigrade, a significant contribution to global cooling. Petit et al. [1990] also linked the dust record in the Vostok [Antarctic] ice- core, with palaeoclimatic trends in the southern hemisphere, and confirmed the work of Danny-Harvey [1988] that a significant increase in dust loading of the atmosphere occurred in the Late-Glacial cold maximum. However they estimated that the maximum volcanic contribution to the ice-core dust levels was never greater than 20% of the total micro-particle content.

This discussion of the construction and application of Eruption Chronologies has been necessary in order to demonstrate that the simple association of an eruption with environmental phenomena elsewhere is insufficient proof of a causal relationship. Both the DVI and VEI were shown to be inadequate both in terms of the number of eruptions recorded and the criteria measured. Even in the twentieth century eruptions have gone unrecorded. Extreme caution must be exercised in any attempt to link the few eruptions identified in prehistoric times with palaeoenvironmental change.

In order to be considered climatically significant, a volcanic eruption must emit a significant volume of acid volatiles. Even where a temporal association between an eruption and an environmental phenomenon elsewhere is demonstrated, if the eruption was not acid-producing a relationship between the two events cannot be assumed or accepted. Ice-core data provides a partial proxy record by which acid emissions and dust output can, in appropriate circumstances, be reconstructed. The correlation of an acid-producing eruption and environmental phenomena may suggest a link between the two events. A further measure of this relationship may be found in the environmental response of trees to volcanic forcing of climate. The record of stress in the dendro-chronological record and acid peaks in the ice-core record may indicate that an historic or prehistoric eruption had an environmental imp