The facts of Eldgja are well established. We know approximately when it happened, where it happened, how much lava, tephra and sulphate was ejected. We have found the tephra in Greenland. We think we know the human impact over much of the northern world, arising from three years of winter. But on other aspects, our understanding is on thinner ice. We don’t know anything about the human impact in Iceland itself. We don’t know what caused the fire, or why it erupted as much as it did (and no more). We don’t know which volcanoes can feed such a fire. And we don’t know when the next one will be.
Let’s first clear up a separate confusion. If we accept shifting the Neem and NGRIP ice cores by 5 years, there is no longer any sulphate peak at 934 AD. The evidence for a long-lasting event came from the need to reconcile the 934 and 939 peaks, seen in different ice cores. Now there is only one peak, lasting 2-3 years, securely linked to Eldgja by the tephra in it. But this is not the full story. The NGRIP core still shows a second, weaker sulphate peak, some years after the main one. In the original dates, this second one approximately lined up with the later peak of the GISP2 core. In the new dates, it doesn’t. So by solving the Eldgja problem, we have created a new, later eruption. What was this?
The VEI7 that wasn’t
The two sulphate peaks in the ice core record are separated by about 7 years. In the revised chronology, the second, weaker sulphate peak occurs around 945 AD. Could this have been a final flourish of Eldgja? It turns out that this is not the case. The ash from the first peak has been securely identified with Eldgja. But the ash in the second peak, found both in the NEEM and NGRIP cores, is different. In particular, it has a higher iron content relative to calcium than found in Iceland. A paper by Sun et al. in 2013 showed that the composition of these fragments was a close match to Changbaishan volcano (also known as Baitoushan or Paektu), on the border between North Korea and China.
Changbaishan has a crater lake called ‘Tianchi’. The name means ‘heavenly lake’, and it is the source of three different rivers. But its formation was far from heavenly, unless from a very upset heavenly creature. The lake formed in a massive eruption, a VEI 7, about a thousand year ago. (The North Korean nuclear test explosions are done very close to this inflating volcano. Military stupidity knows no end.) The VEI 7 classification makes it one of the largest events of the millennium – I have seen it listed as among the 10 worst eruptions of recent history. It is sometimes called the millennium eruption, not because of its size or age, but because some estimates have dated it to 1000 AD.
Amazingly, for such a major eruption in a highly developed and literate region, there are no written records of it. In this, it is just like Eldgja. There are indirect mentions. In Kyoto, an explosion (thunders like a drum) was heard in February 947, and white ash fell in Nara, Japan on 3 Nov 946. In Korea’s capital Kaesong, drum-like thunder was heard in 946. The 14C dating has not given a unambiguous date, but the best determinations also put the eruption around 946, within a few years. The identification of its ash from Greenland now confirms this date, and the second sulphate peak can be identified not with Iceland, but with Changbaishan, a continent away.
It is a funny coincidence that there were two major eruptions from entirely different volcanoes, separated by about the same amount of time as the uncertainty in the dating of the ice cores. Science can be difficult. (For some reason, major eruptions more often happen 7 years apart: volcanoes do a 7-year itch. This happened in the double event of 540, the double event of 1453/1495, and the two big bangs of 1809 and 1815. Coincidence is like the lottery: funny things happen.)
But if Changbaishan occurred in 946/947, why was there no volcanic winter after this? A VEI 7 is massive and all known VEI-7 eruptions have had major effects. It turns out this eruption may have suffered from a numerical error. A VEI 7 eruption is defined as more than 100km3 ejecta. Changbaishan indeed erupted 100-120 km3 in tephra, but the DRE value is much less, 30 to 48 km3 – making it a ‘run of the mill’ VEI 6. The relatively small caldera, 5 km wide, agrees with this assessment. It was big, but not world shattering. That “Tianchi” was a VEI 7 appears volcanological fiction.
The sulphate deposition in Greenland from Changbaishan’s millennium eruption was much less than that of Tambora, and even less than Krakatoa. Those two eruptions were tropical and a northern eruption such as Changbaishan should deposit more sulphate in Greenland, not less. Oppenheimer lists it as ‘only’ M6.8 (about a third smaller than the smallest VEI-7), but even that may be a high estimate. It was a significant eruption: downwind in Japan the ash layer was 10 cm thick. But Changbaishan was too small for major climatic consequences.
This is no excuse for making it the only volcano in the world subject to experimental nuclear detonations!
Feeding the fire
After this Korean diversion, let’s get back to Iceland. There have been four major ‘fires’ in Iceland since the ice age, all four in the East Volcanic Zone. Two were in historic times: Laki and Eldgja. The third was around 8000 yr ago, the Holmsa fire, with a volume of at least 5 km3. Its lava is mostly buried making it hard to accurately measure the volume. Holmsa occurred on the same NE trend where Eldgja also erupted, but the exact source region has not been found and may have been covered by Eldgja lava. The fourth, and largest, was the Thorsja lava flow, 8600 yr ago, estimated at 22 km3. The Thorsja fissure is to the north of Eldgja.
The fact that each of the fires here comes from a different rift shows that the EVZ is a wide, weak zone which can break in numerous places. It is pulled apart, breaks, stitches the break with solidifying magma (the ultimate poly filler), and breaks somewhere else. It doesn’t offer much resistance! Half of all lava erupted in Iceland since 900 has come from here. And this is in spite of the EVZ not having any major volcanoes itself. It is a conduit for volcanoes located elsewhere, and is being fed from its edges: Holmsa and Eldgja came from Katla, Laki came from Grimsvotn, and Thorsja came from Bardarbunga. Each of these three major volcanoes produced one (at least) massive fire in the EVZ. Iceland is truly a land of equal opportunities.
Why doesn’t the EVZ have a big volcano of its own? Perhaps it is too weak to hold down a big enough magma chamber. That it lacks a big magma chamber is shown by the topography: the area is much lower than the raised peaks of Mýrdalsjökull (Katla) and Vatnajökull (Bardarbunga/Grimsvotn). Hot magma has lower density than the surrounding rock, and therefore lifts up the land above it. Just by looking at the topography you can already get an idea where the magma chambers are. (This only works well for liquid rock: solid rock can be out of equilibrium. But it works well for magma.) Gravity reveals what’s hidden below, and it shows that although the EVZ may not be devoid of hot magma, it lacks huge reservoirs.
A matter of gravity
It is an interesting fact that each of these three main eruptions along this rift zone produced similar amounts of lava. Effusive eruptions are fed by a pressurised magma chamber. In the case of Holuhraun, the main force pushing magma to the distant exit was gravity: the exit point was a kilometer lower than the top of the mountain, and the weight of the mountain pushed it out. Could this be true for the three main fires as well?
Let’s estimate the pressure on the magma chambers from the weight of the mountain. The magma chambers are typically 10 by 10 km, with a surface area of 100 km2. The magma chamber carries the weight of the mountain above; the rift next door has no mountain on top of it and carries much less weight. The magma chamber is therefore over-pressurised with respect to the rift: if it can connect to the rift, the magma will flow towards it, just like the communicating vessels from school physics lessons. This is true even if the leak is much deeper than the rift.
The top of the mountain is about 1 km above the rift. The total volume which gives the excess weight is the surface area of the chamber, multiplied by this height, which becomes 100 km3. This is the most magma that can be pushed out by the mountain. In the case of Holuhraun, perhaps a quarter to a third of the magma coming out of Bardarbunga ended up on the surface: the rest was used to fill up the rift. If the same ratio holds for these very large fires, the most that they can erupt becomes about 30km3. The numbers make sense. Eldgja was not far off the largest fissure eruption Iceland can support.
Katla’s caldera is about 140 km2 large and 750 m deep. This gives a volume of 100 km3 (a bit overestimated), consistent with the numbers above. The bottom of the caldera is at about the same height as the Eldgja rift, perhaps not accidentally. Bardarbunga’s caldera covers 80 km2 in area and is 700 m deep: it has a volume of 56 km3. The bottom is at 1100 m, which is a bit higher than its fissure eruption but the caldera may have recovered a bit since its fire. Grimsvotn’s caldera is a bit smaller, at 50 km2. I have not found a value for its volume, but the smaller area would be consistent with the fact that Laki was a smaller eruption than Eldgja or Thorsja. The numbers hold up: it is conceivable that the Icelandic fires are gravity fed.
This implies that these three calderas may have formed in their respective fires. Grimsvotn’s caldera in particular would be very recent, and is the missing hole left from feeding Laki.
Caveats Two disclaimers are needed. First, the fires are not the only way Iceland can form calderas. Individual volcanoes can also have large explosive eruptions, which are pressure driven and can eject vast amounts. But in recent centuries, the major fires have been dominant and the current calderas would be due to the fires. Calderas can form suddenly, but can also disappear rapidly, within centuries, as the magma chamber refills.
The second disclaimer (speculation alert) is on the maximum eruption size. By keeping the rift narrow, it is possible to use less magma to fill the rift so that more is available to be erupted. That could double or triple the erupted volume, from the same amount of magma. But it wouldn’t be a fire: because of the narrower rift (or dyke) it would erupt at a lower eruption rate, effusive but without the flow volume needed to maintain the enormous fountains over an extended fissure. The lower eruption rate would tend to build a shield volcano, slower to grow but potentially reaching a large volume. The EVZ notably lacks shield volcanoes.
Katla’s state of the nation
Katla has a long history of eruptions. Over the past 8400 year, it has erupted 350-400 times, about four times per century. The large majority were explosive: only about 10 effusive eruptions have been identified, and 8 of these were minor. Eruption rates were high between 2000 and 4000 yr ago (when the two largest explosive Katla eruptions occurred), and between 7000 and 8000 yr ago. The modern rate is only half of the average: Katla has gone a bit quiet (this should not be overstated: it is a bit like a teenager with ADHD having a better day.) This better day may have started after Eldgja.
It is worth noting that Katla had a significant eruption around 920, 15-20 yr before Eldgja. There was no quiet period before the fire. Katla ‘fired’ without needing time for any extensive preparations. The eruption came from its normal supply of magma, without the ominous silence that other volcanoes use to signal danger ahead. It happily continues to erupt in its usual regular irregularity, and suddenly wipes out Iceland. It is as predictable as that North Korean – the one trying to detonate a volcano the nuclear way.
Katla’s eruptions invariably melt vast amounts of ice, which comes out as large floods. The jokulhlaups carry several times more sediment than the eruption ejects! Large ones occur twice a century but small ones, caused by geothermal melting, happen every year. Before Eldgja, the jokulhlaups emerged in various, unpredictable directions, but afterwards they have only appeared from the eastern glacier. Something has changed at Katla. The Eldgja rift may provide a channel for the melt water, breaching the caldera, and that still rules the floods.
The reduced eruption rate and the cavernous caldera suggests that Katla is still recovering from Eldgja. It is unlikely to put in an Eldgja repeat performance any time soon.
As I write this, Katla is responding by shaking its insides, clearly protesting against this assessment. Everything Katla does is pre-eruptive, because Katla erupts so often and by definition is always in an pre-eruption state. Even the summer snow melt is pre-eruptive activity. Even a butterfly settling down could set it off. It is the angry young man of volcano-world. If these are Katla’s quiet years, imagine what it was like in the times when it erupted twice as often. But the chance of the current shaking leading up to an Eldgja-II is somewhere between slim and zero. It isn’t yet ready.
Back to the future
There are two remaining questions. Why, when the smaller Laki eruption was so devastating, do we not have historical records of the Eldgja disaster, and second, when will it re-occur?
The first question can only be answered with speculation. It is hard to believe that Eldgja would not have badly affected the Icelandic population. The region most affected, in the south, may have been thinly populated (although the Book of Settlement claims otherwise), or largely illiterate. Similar to Laki, agriculture in large areas of Iceland would have been hard hit and the population declined, although perhaps the food supply was more reliant on the sea than on produce from the land. Eldgja began just after the official end of the 60 year of settlement. Perhaps this period ended because Eldgja had made further immigration unattractive. The lack of written records could be directly due to the economic and human damage. But it really is an unanswered, and important question.
As to the future, we have been very unlucky to have two of the three major fires happen during the brief time Iceland has been occupied. It was bad karma. On average, they have happened only once every 3000 years. There is only a 3% chance of a Laki within the next century. No worries. Unless, that is, if the last 1200 year are the new normal.
Looking deeper, the three main source volcanoes for the EVZ may not yet be able to repeat their performance. The calderas are still too deep for gravity to act its full power. Of course an area offset from the existing caldera could act up, but eruptions do tend to come from the existing calderas, either directly (Grimsvotn) or indirectly by feeding a fissure (Bardarbunga). But there is one question mark. Because this area of Iceland does not have three major volcanoes, it has four. And the fourth is quietly biding her time.
Hekla is the one volcano in the region which could feed a major fire. Will it? Probably not. She may have a fiery temper, she does not have a fiery habit. But do be wary of the Lady.
Eldgja: The End
The search for Eldgja has been a fascinating experience. So little is known, and what is there is scattered over the literature. Pulling it together revealed the holes in our understanding, and isn’t that what research is all about? But a better understanding is important. What Iceland did once it could do again. What Iceland did twice it will do again. Two of the six worst volcanic affairs over the past 1200 years have come from Iceland. To be prepared, we need to know what happened. The next step in the Eldgja saga should be to solve the dating problem: with a precise date we can look for its impacts around the northern hemisphere. And finally, archaeology of affected Viking sites would tell us how it changed their world. It could change ours.
Albert Zijlstra, September 2016
T. Thordarson G. Larsen: Volcanism in Iceland in historical time: Volcano types, eruption styles and eruptive history. Journal of Geodynamics 43 (2007) 118–152
Ash from Changbaishan Millennium eruption recorded in Greenland ice: Implications for determining the eruption’s timing and impact. Chunqing Sun et al. Geophysical Research Letters, 41(2), 694-701. http://pure.qub.ac.uk/portal/files/13968752/Sun_et_al._2014.pdf
M. G. L. Baillie and J. McAneney: Tree ring effects and ice core acidities clarify the volcanic record of the first millennium. Clim. Past, 11, 105–114, 2015. http://www.clim-past.net/11/105/2015/cp-11-105-2015.pdf
Michael Sigl et al. A new bipolar ice core record of volcanism from WAIS Divide and NEEM and implications for climate forcing of the last 2000 years. Journal of Geophysical Research: Atmospheres, 118, 1151–1169, 2013
Michael Sigl et al. Timing and climate forcing of volcanic eruptions for the past 2,500 years. Nature, 523,543–549 2015
Gregory Zielinski et al., Evidence of the Eldgjá (Iceland) eruption in the GISP2 Greenland ice core: relationship to eruption processes and climatic conditions in the tenth century. The Holocene, 5, pp 129-145 (1995) D. McCarthy & A. Been: An evaluation of astronomical observations on the Irish Annals. Vistas in Astronomy Vol. 41, No. 1, pp. 117-138, 1997
Magdalena Schmid et al. Tephra isochrons and chronologies of colonisation. Quaternary Geochronology (2016) T. Thordarson et al. New estimates of sulfur degassing and atmospheric mass-loading by the 934 AD Eldgjá eruption, Iceland. Journal of Volcanology and Geothermal Research, 108, 33–54 (2001).
T. Thordarson et al. New estimates of sulfur degassing and atmospheric mass-loading by the 934 AD Eldgjá eruption, Iceland. Journal of Volcanology and Geothermal Research, 108, 33–54 (2001).