Iceland is the most volcanic place in the world. Explosive eruptions, from many different volcanoes, are common, unpredictable and, when seen from afar, exciting. But don’t be fooled. These volcanic explosions are small fry, and mostly harmless. Less common but far more devastating are the other type, responsible for 80% of all Icelandic lava and 99% of its volcanic casualties: Iceland’s fissure eruptions.
Fissure eruptions occur along a linear fault rather than at a central volcano. Elsewhere in the world they are known as Icelandic eruptions, but in Iceland they are known as ‘fires’. A vigorous fissure eruption produces curtains of fire along the fissure – hence the name. Over the past 1200 years, there have been 14 fires in Iceland. Four of these were in the past 100 years, an excess which suggests there may have been more and older records are not fully complete. The most recent fires were Holuhraun (2014) which produced over 1 km3 of lava, drawn from below Bardarbunga 40 km away, Krafla (1975-1984), Surtsey (1963-1967), and Askja (1921-1929). Holuhraun in particular was a large, impressive fire. Bigger by far was the Laki eruption (1783-1784), also known as the Skafta fires (Skaftareldar, to be precise), which produced 15 km3 (DRE), drawn from beneath Grimsvotn. Laki was devastating and destructive: 78% of all horses on Iceland were killed, and 50% of cattle. The human population declined by 22%, there were significant fatalities from air pollution as far away as the UK, and Laki caused a winter so severe that as many as 1 million people may have perished from cold and starvation. It was a fire from hell.
And Laki was not even the largest fire on record. The Eldgja eruption which began in 934 AD was bigger, at almost 20 km3. It happened not longer after the settlement of Iceland began – the Vikings initiation to Iceland’s volcanoes was a baptism by fire.
The name is self-explanatory: ‘Eld’ mean fire, and ‘gja’ means fissure. But don’t let this name fool you into thinking this is what was seen. Although Iceland was already well occupied at the time (the population around 930 is estimated at 30,000), no written or oral records exist of the eruption. It is as if no one noticed it. The fires would have been visible from many places in the south of Iceland. Soon those areas would be covered by thick tephra, clouds of sulphuric acid, and overrun by floods of lava and water. How could it not have been noticed? Were all early reporters wiped out?
The duration of Eldgja has been estimated as 3 years (Hammer), up to 6 years (Zielinski) or even 8 years (Strothers). There are claims that weather was disrupted over Europe and Asia for as long as 9 years. But can this really be correct, if Laki only managed two years of mayhem? The story of Eldgja is far from complete: what are the facts – and what is fiction?
Let’s first look at the early Icelanders who themselves experienced Eldgja. The oldest civil records of Iceland are in the Book of Settlement and the Book of the Icelanders, both written down during the 12th century, mostly by the same person. They describe how the Vikings first settled Iceland in 874 AD or 870 AD (the two books don’t quite agree). 870 AD sounds more plausible: the winter of 873/874 was severe in Europe with significant death rates (reported as high as 30% in Mainz) which is not ideal for starting a colonization. On arrival, the Vikings found evidence of earlier occupation by Irish monks, but it is not clear whether these monks were still there at the time or had already left. The period of settlement lasted for 60 years, by which time the population may have been as high as 30,000. The Icelandic parliament, the Althingi, started either 930 or 934. The Books give names and fragments of background on many early settlers.
Archaeology confirms this sequence. There is a convenient tephra layer in Iceland dated to 877, and remnants of a few Viking sites have been found below this, all on the south-west coast. This favours the earlier date of 870 for the first settlers. Between the two tephra layers of 877 and Eldgja there are many Viking sites, particularly on the north side of Iceland. Many sites are well in-land, suggesting the coast had already filled up. Iceland was much more forested and vegetated than it is now. The Vikings settled on farmsteads, but the soils were often thin and quickly eroded. Farming became marginal and the carrying capacity of Iceland was reached quickly. Through various ups and downs (including the black death), by 1700 the population was still only 50,000. After Laki it was down to 40,000, not much more than in 930!
Eldgja is thought to have started in 934 and to have lasted to perhaps 941 (but I will come back to dates later). The eruption eventually covered 840 km2. At least 6 km3 of tephra was deposited over the country (the actual amount may have been larger: some will have ended up at sea). The lava volume is not fully known, as part has since been covered by sediments. On the western side, sediments from later jokulhlaups from Katla have covered some of the lava flows under a depth of 10 meter. The most recent estimate of the size of the eruption is 19.8 km3 (DRE), which includes the tephra. For comparison, the 1783-1784 Laki eruption was 14.7 km3. The Laki flows covered some of the Eldgja ones, which is one way to usurp your predecessor. But although both eruptions occurred close to each other, in the East Volcanic Zone (EVZ), they were fed by volcanoes on opposite sides of the zone.
The eruption came from within a 200-meter deep rift which ran north-east from the Katla glacier. Four separate rift segments are recognized. The west (or southern) rift, closest to Katla, extends 9 km from the ice cap and ends at the crater of Raudibotn. The rift extends underneath the icecap, perhaps by the full 15 km to the caldera. A series of spatter cones have formed in the west rift. About half the Eldgja lava was erupted from here. Where the west rift ends, a section of graben survives, which continues for 3 km. This section ends at the mountain Svartahnjuksfjoll and did not erupt. Here the central rift begins: it runs for 9 km along the western margin of the Alftavatn valley. The craters along this portion of the fissure consist of a series of ‘gjas’ which don’t quite connect. At the northern end of the central rift, the fissure jumps 1.5 km east where the eastern fissure begins, also called Eldgja proper. It runs for for 8.5 km, terminating in the mountain Gjatindur. The fissure includes the popular tourist attraction Ofaerufoss, with a spectacular waterfall.
Another graben follows, until it reaches the Skafta river. Here the northern fissure begins, with a row of spatter cones extending 19 km. The lava flows from this section were later covered by the Skafta fires, which fissured parallel to but a little south of the northern rift. It was only recognized late that the northern rift was part of Eldgja. The total length of the fissure is 57 km from the edge of the glacier, or over 70 km if, optimistically, counting from the caldera.
Before the Eldgja eruption, the entire rift was a graben, which had existed already for some time. The lava filled the graben and flowed out mainly following river valleys, but fragments of the graben have survived. There are three main lava fields. Each followed a valley or river to towards the coast, and spread out on the coastal plain. The largest is the Alftaver flow which came from the western rift and accounts for half of Eldgja’s lava. The Landbrot and Medalland flows account for most of the rest: they came from the central and eastern rift. Lava flows from the northern rift are much smaller, estimated at 0.5 km3: these have largely been covered by Laki. One hyaloclastite flow has been identified next to the Katla glacier, from an underwater (glacial lake) eruption.
The lavas are a type of basalt rich in titanium, as typical for Katla. But there are complications: the northern and to some degree the eastern rift seem to contain another, tholeiitic component as well. It has been suggested that at the northern end, there was an inflow (or prefill) of this lighter magma from Grimsvotn, and that the eruption in this location was a 50/50 mix of Katla and Grimsvotn magma. That would be very unusual: eruptions may draw from different regions in a magma chamber, but not normally from different volcanoes. The push from one volcano will override that of the other. But perhaps it can happen in a spreading ridge where the lava is sucked in rather than pushed in.
The eruption sequence is only partly known. From the fact that ash was found in the Greenland ice cap, we can deduce that there was an explosive event, most likely underneath the ice cap, perhaps within the caldera where most of Katla’s eruptions happen. The Eldgja tephra in Iceland shows over 30 distinct layers, over at least 8 separate episodes: the eruption was not continuous but there were a different events at different times. The lowest tephra layers are thickest towards the Katla glacier. This shows that the eruption began there. Later, when the rift opened, the eruptions migrated north east.
Each of the rift eruptions would have started with strong earthquakes, typically M5. Laki showed such earthquakes throughout the 8-month eruption. Each new eruption was probably phreato-magmatic to begin with (the top layer of the Icelandic crust tends to be quite wet and there may have been lakes in the graben): the tephra show alternating layers of phreato-magmatic and pure magmatic tephra. Than the fire fountains begin. For Laki, the fountains reached a kilometer high; Eldgja may well have shown similar heights. The glow would have been visible over all of southern Iceland, and the fountains seen on the horizon. What a view it would have been, magnificent and terrifying. And during the months and years of the eruption, this may have happened over 30 times.
The fountains build up cones, and lava lakes develop inside the cones. The lakes greatly reduce the height of the fountains (because the exit hole becomes larger). The eruption remains just as fast but is less visible. Later, as more magma has been pushed out, the internal pressure reduces and the flows begins to wane. Eventually it becomes so slow that the magma underground has time to solidify, and now the eruption stops. This is how Holuhraun ended. But underground, with the exit blocked, the pressure from behind begins to increase again. The magma may break through in a new place (this did not happen in Holuhraun -there wasn’t enough magma left). You expect a waning eruption to work its way back to the origin, with new break-outs happening closer to the actual volcano. But Eldgja appears to have migrated the other way, stepwise away from the origin. Perhaps the eruption was not solely controlled by the magma supply. A migrating rifting event may have played a part.
Eldgja, like Laki, erupted huge amounts of sulphur, about 220 Mt of SO2. This is much higher even than Tambora which produced about 80 Mt. Much of the environmental and health impact comes from this.
Finally, the eruption subsided and Iceland was left in peace. But much of the region was devastated. It has been argued (for instance by Mathias Nordvig) (but remains speculative) that the Eldgja eruption is behind the saga of Ragnarök – the twilights of the gods.
To boldly flow
Before the eruption, the land was very different. It had been sufficiently vegetated to allow for several settlements, but the soil was thin and easily eroded, and early settlements were apparently quickly abandoned. The eruption changed the landscape and moved the coast line.
Not all flows were of lava. Katla eruptions invariably lead to water floods, jokulhlaups which appear from underneath the glacier. The jokulhlaups travel typically at 20 km/h, and without warning it is difficult to outrun or survive them. The deposits are eroded rock from the mountain underneath the glacier and soil from the upstream flows. Jokulhaups left substantial deposits after or during Eldgja around the ice cap, on the western edge of the lava flows. Katla’s eruptions typically deposit more volume from the jokulhaups than from the eruption itself, although for the Eldgja eruption the lava flows were more voluminous.
Furthermore, the tephra ‘floods’ would also have been deeply damaging. An area of 2600 km2 was covered under more than 10 cm of tephra, and 600 km2 under more than 1 meter, about the same area as was covered by lava. The tephra thickness dropped below 0.5 cm only beyond 100 km, which is twice as far as for Laki. Whereas the lava and jokulhaups flow downhill, tephra goes where the wind blows. Even a centimeter of tephra is destructive to agriculture and many areas would have many years to recover.
Coming up So far, we have focussed on the facts, the things we know. The next part will talk about less certain aspects: the date, the volcanic winter, and how the fire was fed. And finally, could this happen again?