Hekla is the most mysterious of Iceland’s many volcanoes. Its brooding summit overlooks the broken plains 800 meter below as if it were an English Lord (or perhaps Lady) of the Manor. The fiefdom looks bare and uninviting, but that is not purely Hekla’s fault: once this was dense forest, but it was cut down by the villeins. But Hekla never was a harmless eccentric lord; it is vengeful, evicting its cottars and killing its serfs. 13% of Iceland’s historic eruptions have come from Hekla. So has 18% of Iceland’s lava output (or 6% of the total on-land lava output in the world), and 8% of its tephra. Pity the poor souls whose lord was Hekla!
Carl wrote a compelling post on the unpredictability of Hekla, where the size, progression and spacing of each eruption seems absurdly different. Hekla’s explosive eruptions can be severe: according to one report, a 20-kg lava bomb was shot out to a distance of 32 km from the mountain. It could have brought down an airplane, had airplanes existed at the time. The effusive eruptions are equally impressive. And the tephra is as dangerous as the lava: Hekla’s many hazards include widespread poisoning by fluorine.
The mountain has the reputation of erupting almost without warning. After the earthquakes begin to be felt, the eruption may only be an hour away, bad news if you are on the summit and need several hours to get to safety. There is an impressive description of the eruption of 1845:
On the night of 1 September, the dwellers in its neighbourhood were terrified by a fearful underground groaning, which continued till mid-day on the 2nd. Then, with a tremendous crash, there were formed in the sides of the cone two large openings, whence there gushed torrents of lava, which flowed down two gorges on the flanks of the mountain. The whole summit was enveloped in clouds of vapour and volcanic dust. The neighbouring rivers became so hot as to kill the fish, and the sheep fled in terror from the adjoining heaths, some being burnt before they could escape. On the night of 15 September, two new openings were formed — one on the eastern, and the other on the southern slope — from both of which lava was discharged for twenty-two hours. It flowed to a distance of upwards of twenty miles, killing many cattle and destroying a large tract of pasturage. Twelve miles from the crater, the lava-stream was between forty and fifty feet deep and nearly a mile in width. On 12 October a fresh torrent of lava burst forth, and heaped up another similar mass. The mountain continued in a state of activity up to April 1846; then it rested for a while, and began again in the following month of October. Since then, however, it has enjoyed repose. The effects of these eruptions were disastrous. The whole island was strewn with volcanic ash, which, where they did not smother the grass outright, gave it a poisonous taint. The cattle that ate of it were attacked by a murrain of which great numbers died. The ice and snow, which had gathered about the mountain for a long period of time, were wholly melted by the heat. Masses pumice weighing nearly half a ton were thrown to a distance of between four and five miles
— Anonymous, 1872
Hekla is believed to be young. The tephra layers found in Iceland and Europe which are ascribed to Hekla are all from the Holocene. The only real evidence of volcanic activity from before the Holocene comes from some fissure lavas on the flanks which are dated to the Pleistocene. (The area is heavily eroded and some evidence of older activity may have been lost.) Befitting its suspected youth, it is still growing: the 1947 eruption added 50 meters to the height, to 1501 meter above sea level in total. Erosion has since removed 10 meters. At this net rate, the volcano might only be 1000 years old. In reality it is rather older, but still less than 10,000 years, a teenager with a volcanic temper. And like any teenager, it hasn’t settled yet into predictable behaviour: over the months and years it continuously changes.
Its youth is also evident from the local motion of the ground. Iceland as a whole is rising, a consequence (still) of the adjustment after the end of the ice age. The almost (but not quite) complete ice cover, 500 meter thick, pressed down on the crust. When the ice melted, the crust began to rise up, currently at rates up to 20 mm per year. (Since 2003 this rate has been increasing, which is attributed to the melting of the remaining glaciers.) But around Hekla, the crust is not rising as fast. This is due to the weight of Hekla, to which the crust has not yet had time to fully adjust. The lack of full adjustment shows that the bulk of Hekla is younger than the ice age. Over the past 1000 years, some 8 km3 of lava has been added to the surrounding plains and the bulk of the growing mountain.
Hekla is a typical, pretty stratovolcano. But this is only when viewed from the southwest. Look from the east and it becomes a ridge volcano – an upturned boat. The volcano is in fact quite elongated, and clearly follows a linear fissure rather than being tied to a singular point. Compared to the surrounding volcanoes, the fissure is quite short. While Katla feeds eruptions tens of kilometers away, the fissure of Hekla only reaches 3-4 kilometer from the summit. But it does have a summit, and most of its eruptions start there. Hekla is a combination of a normal volcano and a crater row, a combination it shares with Mauna Loa.
It is not the only such volcano in the area. There are some 40 mini ridge-volcanoes surrounding Hekla. They are called mobergs (or table mountains), and are shaped like Hekla and oriented in the same direction, but much smaller. They are a bit older, and date from the ice age. The bottom layers of the mobergs are pillow lavas, which were erupted under water: they were immersed in melt water which had collected under Iceland’s ice cap. Higher layers were erupted in air. The composition of the lavas is similar to Hekla. This region has had a magma supply since at least 20,000 years, although perhaps not a single magma chamber. Hekla itself became dominant only in younger times. It may have started as just another moberg, but it outgrew the rest, perhaps because by that time a single magma chamber had formed.
Hekla is a frequent erupter. As Carl writes, the pattern has changed over the years, and where previously it erupted something like twice per century, in recent years it has been as often as once per decade. There have been 18 known summit eruptions since 1104, and a few more eruptions solely outside of the summit. Combined, the eruptions eject around 1 km3 (DRE) per century, in lava and tephra. The 1947 eruption did this much in one go and the 1104 eruption may have put out even more. Much of Iceland frequently receives a dusting of Hekla ash, laced with a generous seasoning of fluorine, deadly for cattle but good for their teeth.
But in spite of its significance, Hekla is a poorly understood volcano. What caused Hekla to form? What will it become? And where did it get its first magma from? Even the depth of the magma chamber is uncertain. The science of Hekla progresses slowly, bit by bit, eruption by eruption. Let’s see whether we can get to the bottom.
Eruptions and lava
Hekla has two eruption modes. The first mode is typified by plinian or sub-plinian events from the summit, which produce dacite and rhyolite. There have been at least five very large such events, the oldest around 5000BC and the most recent in 1104 AD. Each ejected 0.5-2 km3 DRE, explosively. The second eruption mode consists of effusive fissure eruptions, producing copious amounts of basaltic andesite.
In practice, eruptions often mix the two types. They may start as explosive events from the summit, followed by effusive events along the 8 km long fissure through the summit, or along fissure swarms parallel to this.
The rhyolitic component has around 65% SiO2 by weight, and the andesitic component about 55%. Either these two come from different magma reservoirs, or the main reservoir contains differentiated magma. No known Hekla lava has SiO2 below 54%.
Since 1947, only the andesitic component has erupted. The lavas have become much more uniform in composition, and the recent composition is near the bottom end of the range of SiO2 fraction seen in earlier times. There is an evident correlation, where the initial SiO2 fraction of an eruption is higher if there has been a longer repose time. That suggest the magma is evolving (or recovering) during the quiet periods between eruptions.
During an eruption, the SiO2 fraction smoothly decreases as the eruption progresses. This shows that the magma reservoir is stratified. The magma with the highest SiO2 concentration, which is also the lightest, floats on top. As the eruption continues, it delves deeper into the magma reservoir and so mines the heavier material. But only to a limit, as Hekla does not erupt lavas with SiO2 content below 54%. Other volcanoes in the area do erupt magmas with less SiO2 than this: this limit is unique to Hekla. It suggests that the deeper, denser material in the magma chamber, with such a composition, is too dense to erupt. The buoyancy is not enough to make it move. After the eruption ends, the top layer slowly evolves back towards higher SiO2 content.
To keep the magma chamber so well stratified, it must be rather quiescent. There is no major heat source that causes the magma to turn over: there is no lava-lamp effect. Thus, Hekla is not driven by rising heat from below: its location must be an accident of the local geography, not driven by local mantle heat. Of course something must be melting the magma, but either that is a heat-free (adiabatic) process (decompression melt), the melt occurs elsewhere and percolates into the Hekla chamber, or the melting happened in the past but not now. The lack of heating explains the SiO2 limit of 54%: the buoyancy is insufficient because the temperature does not increase enough with depth. A shallow temperature gradient can prevent lava lamp (or rather, magma lamp) motion.
So now we know what the magma chamber must look like. But where is it? This is a question on which volcanologists almost come to blows, or at least as close as is possible within the language of science. Calling someone else’s model ‘excluded’ or ‘impossible’ is the science equivalent of holding Hamlet’s dagger before their eyes, and putting those words into a scientific paper amounts to a declaration of war. For scientists too, Hekla brings out the worst in us. It is the portent of hell.
Geirsson et al. argue for a deep chamber, below the crust, located at a depth of 24 km. They use GPS stations around Hekla which measure the local movement of the land. This is a complex problem, as the GPS motions incorporate many different movements: the local transform, the spreading ridge, the isostatic rebound, and any motion caused by local earthquakes. Subtracting all these, they find that the area around Hekla shows expansion away from the volcano. Hekla is getting bigger.
To fit the expansion pattern, Geirsson et al. show that a very deep spherical or ellipsoidal, inflating chamber is needed. This model does not reproduce all velocities north and west of Hekla. The fit improves with a second chamber at 3-8 km depth, displaced from Hekla by 6-11 km to the west. This displacement is disputable, and the model is certainly not perfect.
In their GPS models with a deep and a shallow magma chambers, both chambers are inflating but at different rates. The shallow chamber is growing at 0.0004–0.0014 km3/yr. The deep chamber inflates much faster, at 0.02 km3/yr or so. To put this into context, the rate of inflation even of the shallow chamber could fill the entire volume of Hekla within 10,000 years.
An entirely different way of finding the location of the magma is by measuring the strain and tilt along a baseline on the flanks of Hekla. Such measurements during the 2000 eruption put the erupting chamber at a depth of 11 km (most recently by Hautmann et al 2017): this doesn’t fit either of the two GPS-derived reservoirs but does fit the earthquake sequence during the eruption, which started at 9 km depth. The two models appear incompatible.
But looking at it with a critical eye, there are complications with both results. The tilt measurements are very sensitive to what happens at the summit of Hekla, and the movement there is dominated by shallow inflation and is less affected by what happens deeper. So it can (and does) show that there is shallow magma, and convincingly shows that there isn’t any even shallower activity, but it does not rule out the presence of deeper chambers. It is also assumed that the chamber is directly underneath Hekla: if displaced to the west, the reservoir could be shallower than they derive (but not deeper). The GPS models, in contrast, measure a large-scale pattern and this is very sensitive to what happens at deeper levels. So there may be truth in both results: they are not as mutually exclusive as may seem.
Sturkell et al (2013) take an intermediate position. They also find that the magma chamber is around 10 km deep, but are more conciliatory regarding the very deep chamber, pointing out that changing the shape of the chamber from spherical to a sill or a pipe will change the derived depth for both models. For instance, one of Geirsson’s models makes the deep chamber wedge shaped, with the top end at 11 km.
It should also be noted that the strain and tilt measurements were analysed for the time of the 1991 and 2000 eruptions, and show especially where the magma moved during the eruptions. Deeper magma, if it lacks buoyancy, may not be responding well to the eruptions and could have been missed. For instance, both a deep chamber at the Moho interface, and a shallow chamber underneath Hekla may be present, but with the eruptions driven mainly from the latter. And the strain measurements also suggested that during the 2000 eruption, the reservoir at 11 km was losing less volume than was erupted from the top, and this requires magma replacement. It does not say from where the magma came, but it may hint at a deeper truth.
In the end, both measurements revealed something, but like the blind man and the elephant, each provided only part of the full picture. We can conclude that magma is very likely present at a depth of 9-11 km, and there may be a significant amount at much greater depth but the details of this region are not well understood.
There is a third method to measure the depth of the eruptive magma. This comes from analysis of the tephra composition of the H-3 eruption, the largest Hekla event during the Holocene. The composition of the tephra shows that the magma that formed the tephra had been stored in a chamber at fairly low pressure and temperature. Laboratory experiments were done by Weber & Castro (2017) to find the precise numbers, and these indicated that the magma storage chamber was some 5 km deep, with a temperature of 850 C, with the magma saturated in water. The magma itself could not have melted under these conditions: it must have come from further down. But there is no evidence for a chamber at 5-km depth in the current strain/tilt-derived models: the tilt would have been very sensitive to any activity in such a shallow location, but none was seen. (Some older work did conclude the presence of this chamber during the recent eruptions, but this was due to a latency error in the BUR station. Once this was removed, the chamber was gone.) Perhaps Hekla has changed, and the shallowest chamber did not survive the Holocene VEI-5 eruptions. Or it is a Chamber of Secrets, still there but hidden from muggle view.
Why is Hekla so rich in fluorine? It is not unique in this: fluorine lavas are more common in South Iceland, for instance in Laki’s lavas. Fluorine is not that rare in the crust (Derbyshire, close to my home, is known for it) but it is less abundant in the mantle. It gets into the magma mainly by melting of the crust, and it tends to increase together with the SiO2 content.
The area around Hekla is far from faultless. It is where the South Iceland Seismic Zone, a transform fault, intersects the Eastern Volcanic Zone, a spreading fault. Hekla is not quite on this intersection (that point falls within the nearby Torfajökull caldera): it is closer to the SISZ than to the EVZ. The EVZ is not so much a single line but a broader zone, in which parallel faults can develop. Even though Hekla is still not within this zone, Hekla’s fissure rift extends in the same general direction as those spreading faults, as shown by its elongation in the same northeast‐southwest direction. But that is not unique: the SISZ is itself crossed by numerous southwest-northeast trending faults. They show the inherent, inborn weakness of this land, and this weakness is shared by Hekla.
The locking depth of the SISZ is around 15 km. Below that, the rocks are ductile and little or no stress builds up. But above the locking depth, the motion of the transform fault and the bookshelf SE-NW faults cause considerable stress. Earthquakes up to M7 can occur here.
The combination of faults makes Hekla a funny beast. Its behaviour is governed by the transform fault, and that is a bit of a rarity as transform faults do not produce magma in the way that spreading ridges do. Neither do slip-strike faults magma make. Perhaps the multiple faults generated a weakness in the brittle crust where magma could accumulate, creating a catchment for magma created elsewhere. The EVZ itself has no currently active volcanoes at this location (nearby Torfajökull is dying). This suggests that it may still be the ultimate origin of Hekla’s magma, where the magma is transported sideways.
Hekla is related to two different faults, neither of which is a spreading ridge. That is good news for the magma that it does accumulate. In a spreading ridge, the majority of the magma creates new crust underground, filling the gap while the two sides separate. Only a small fraction ever makes it to the surface. Below Hekla, no new crust is needed and all the magma can be channelled up to create the mountain. The growth rate of the underground magma chamber was measured from the GPS models at 0.02 km3 per year. This is fairly similar to the long-term average eruption rate, so most of the magma gets to the top. Only about 1-2% of all magma production in Iceland comes to Hekla, but it erupts a much larger fraction of its magma than is the case elsewhere. Hekla uses its magma well, and that is the reason its lava output can compete with that of the big beasts of the EVZ, Bardarbunga and Katla.
Hekla has the reputation that it gives very little warning from precursor earthquakes before it erupts. It is a passive-aggressive volcano, and seismically rather quiet. Between eruptions, there are occasional weak earthquakes but these are very similar to those of the SISZ and probably just come from the stress related to this. The volcanic silence is almost ominous.
During an eruption, the earthquakes extend from the surface to 9km depth. There is a lack of activity further down. This is another reason to put the top of the active magma chamber, which feeds the eruption, at some 10 km depth. The very limited earthquake activity prior to an eruption indicates that there also is an almost open channel between the magma chamber and the summit: once magma begins to move up, it meets little resistance.
This much-advertised silence is a recent feature. Before the 1845 eruption, described above, there was a much longer sequence of audible earthquakes, lasting 12 hours or more. It appears that at that time, the conduit to the surface was not as open, or was solidified, and the rising magma had to break through. The current silent behaviour may well be related to the much higher frequency of eruptions, which is keeping the conduit hot and ductile. A 50-meter wide conduit can take more than a decade to solidify. But now that it is getting close to 20 years since the last eruption, perhaps the next one will not be as silent.
For both the 1991 and the 2000 eruption, the eruption sequence appeared to be as follows. About 30 min before the surface eruption, magma began to rise from the storage chamber through a cylindrical conduit (50 m diameter). Near the surface (within 700 meters) a lateral dike began to form. At this point the (sub-)plinian eruption started. The dike continued to extend away from the summit for another hour, while the rising magma took another four hours to fill it. A Strombolian eruption burst out along the dike, with effusive lava flows coming down the flanks of Hekla. The eruption decreased in intensity over a period of order 100 days. (‘Of order’ means between 10 and 1000 days.) The dike that ran along the fissure remained close to the surface: it was fed purely through the narrow conduit towards the summit.
Each eruption begins with the most evolved magma that is present, but it quickly delves into lower levels in the magma chamber which have lower SiO2 fractions and by the time the dike bursts, it is filled with the lower-SiO2 magma.
As mentioned above, the prehistoric tephra appeared to have come from a shallower magma chamber, at 5 km. The current eruptions are sourced deeper. Was the shallow chamber destroyed in the earlier eruptions? Did that happen in 1104? In 1845? Or 1947? Did it never exist? The work that found evidence for the shallow origin seems strong, with laboratory experiments to show under what conditions the tephra formed. Perhaps Hekla does change, and one should be cautious with predicting the events of the next eruption from the previous one.
What determines the length of time between eruptions? The magma supply rate seems quite constant. Hekla has erupted around 1 km3 per century for as long as it has existed, and that rate is not far from the current inflation rate of the deep chamber. Most likely, the blame for the dead times lies with the varying strength of the rock lid and conduit above the magma chamber. At the moment (or at least over the past half century), it is weak and a little excess pressure (estimated at 10-12 MPa) suffices to break through. But the longer Hekla waits, the more this lid cools, and the stiffer the rock gets. As Hekla waits, it loses the opportunity, and it will have to wait longer. But after half a century, the rock has reached its final stiffness, while the volatile pressure continues to increase. Eventually the lid will break: in the end, the resistance is futile. The longer the lid holds, the larger and more explosive the eruption will be.
At the moment we have had close to 20 years waiting time. This is longer than for the last few eruptions but still much shorter than was usual in the more distant past. If the next eruption is soon, it will not be particularly large. And although it may start explosively, the main event should be effusive. Or perhaps we have passed the point of no return, and the next eruption will not be until 2050.
The reasons for Hekla
The unusual lava composition (for Iceland) shows that Hekla has its own magma supply. It is not sharing a magma chamber with any other Iceland volcano. About 1% of all magma generated underneath Iceland goes towards Hekla. So why was Hekla not there before the Holocene? Where did the magma go before the era of Hekla? How could Hekla develop so quickly, and what started it?
The mobergs show that there already was a magma conduit, but to the general area, not a single point. But at some point, the conduit became localized and the eruption rate increased dramatically. The most likely cause for Hekla’s birth is that the southwest-northeast fault underneath was forced open, either by spreading or by rotation, allowing a magma chamber to develop. What would have caused this? A plausible cause is the isostatic rebound after the melting of the deep icecaps. The loss of weight of Iceland’s ice caused the land to rise up, and this rise was largest here, close to but not at the highlands and major mountains which kept their ice longer. At the same time, the weight loss allowed for decompression melt, and this added to the pressure from below. Once the fault opened, magma had a way in and the building of the mountain began.
Where is the current new magma coming from? Some may come from the EVZ, some from the on-going decompression, and some may come from the transform fault. There are quite a few possible places! The relatively cool magma does indicate that there is no major heat source directly underneath Hekla. However, the continuing inflation shows that the magma accumulation has not ceased and therefore that the source is still active. The location of Hekla strongly suggests that the EVZ has some role in its magma supply.
If the picture is correct, Hekla may not be a long-lived volcano. Once the ice is all gone and the rebound complete, perhaps the fault will close and the magma conduit will be cut off. In another 10,000 years, Hekla could be history, perhaps with a final caldera-forming eruption. But it could also take off and grow, eventually pull in the EVZ itself. The land here is weak, and the break that defines the spreading ridge can easily change location. It might end up moving east, to the break provided by Hekla, thus providing a lasting magma supply and allowing Hekla to grow to the size of Bardarbunga. Or it may be too close to the SISZ for that to happen. Hekla’s heritage is hard to predict.
Hekla may be a child of Torfajökull. The latter is old, nearly dead, rhyolitic, and is a large caldera with little else. Hekla is young, vibrant, (currently) andesitic, and is a stratovolcano. The relation must be obvious! (Ok, not really.) But they are similar in their location, near the SISZ/EVZ intersection. The actual intersection lies within the caldera of Torfajökull, while Hekla is closer to the SISZ than to the EVZ, although close to both. If Hekla is like Torfajökull, it could live long and prosper.
I think that Torfajökull could have started out like Hekla did. But that was long ago, and over time its magma chamber became stale and rhyolitic. Recently, the deep magma supply ceased or became diverted to Hekla, and Torfajökull has been left dry (and ice free). Taking the image forward, we can imagine Hekla continuing to build and grow, becoming a true strato-volcano. As its rift grows longer, lava outflows start to move into the empty lands northeast, and into the greener land to the south-west. Explosive eruptions keep the top of the mountain flat. And one day, the fissure opens wider than before and lava gushes out in much larger amounts, in a true Icelandic fire eruption. The immense lava flow moves towards the coastal plain, filling the Thorsa river valley. The emptying of the mountain gives Hekla a large caldera. It refills, but it happens again. And again. But eventually, the magma supply slows down, and the magma grows old. Rhyolitic explosions now tear Hekla apart. Slowly, the explosions diminish and Hekla periods of dormancy grow longer. It is not yet extinct, but getting there. And finally, a daughter of Hekla grows up, 20 or 30 km away. But that is a long way away, in an unknown future.
Anyone wanting to visit Hekla should be aware. It lords over its possession, and battles any would-be occupants. This small mountain is striving for greatness. Will it achieve this? That is the biggest unknown of all. If it does, it will not be a big friendly giant: Hekla will be a king (queen?) Canute laying waste to the Viking world. Fear the day this Lord of the Manor claims Iceland’s throne.
Albert, April 2018
For current Hekla data, see the VC compilation
The following papers are used in the post above:
Volcano deformation at active plate boundaries: Deep magma accumulation at Hekla volcano and plate boundary deformation in south Iceland, Geirsson et al. 2012. Journal of Geophysical Research, 117, B11409
Phase petrology reveals shallow magma storage prior to large explosive silicic eruptions at Hekla volcano, Iceland. Weber & Castro 2017. Earth and Planetary Science Letters. 466, 168-180
Magma buoyancy and volatile ascent driving autocycliceruptivity at Hekla Volcano (Iceland).
Hautmann et al. (2017) Geochemistry, Geophysics, Geosystems, 18, 3517–3529
New insights into volcanic activity from strain and other deformation data for the Hekla 2000 eruption. Sturkell et al. (2013) Journal of Volcanology and Geothermal Research, 256, 78–86