Remodelling Hekla – A 1947 commemorative article series

Hekla, eldgos 1991. Loftmynd
Hekla volcanic eruption 1991. Aerial viewing north. Photographer unknown.

The point of this series of articles is to propose an alternative model of dynamics, driving forces and magma fractionation at Hekla. It is thus not utilizing the standard volcanological model of how a mantleplume stratovolcano function.

5 years ago, I started to have grave misgivings about how we interpret and model Hekla, as such I started to doubt the foundation of our current understanding of this volcano. The various standard models for how volcanism functions at volcanoes has time after time proven to be correct and/or useful as tools for understanding and prediction, so completely disregarding those models for a particular volcano is not something easily undertaken.

The problem is that Hekla is a unique volcano and that Hekla is the only volcano on earth in its class. Trying to use models of understanding from other classes of volcanoes is about as fruitful as driving square pegs into a round hole, it will not fit.

I will here use an analogy to describe the problem a bit further. Imagine that you are Isambard Kingdom Brunel and that you are sitting in a quaint village in 1850 draining a pint. You are the world leading expert on all forms and modus of transportation. Regardless if it is horse carriages, steam trains or great steam ships, you are the world leading expert. Suddenly Jeremy Clarkson tears down the street in a Bugatti Veyron and you are left dumbstruck about what you have seen.

You immediately set forth utilizing your great knowledge to try to understand what you have seen. It is obviously not a ship, so that knowledge is useless to you. There is no railroad track so it was not a train, nor was there horses involved as evidenced by the lack of piles of horse dung.

In the end, you concluded that what you saw was a steam driven horse carriage. You of course must disregard that there was no huge steam cloud and no sulphurous smell from bituminous coal, but those are just details that are inexplicable to you, the great Isambard Kingdom Brunel. After concluding this, you happily order your next pint and start drawing a great bridge to span the Victorian age into the future.

Only problem is that your great knowledge has led you astray, it was first of all a car, and second of all you have not explained how Jeremy Clarkson ended up 177 years into the past without driving a DeLorean.

The same thing applies to Hekla, a huge mass of correct knowledge has been applied to a volcano that is not congruent to the particular volcano. Hekla is alternately modelled as a stratovolcano or as a fissure row and all attempts to understand the volcano has utilized those models of understanding.

Whatever Hekla is, it is not a stratovolcano in any common usage of the word. Foremost because it is indeed a fissure, nor does it follow the usual stratovolcano shape. Furthermore, as we shall see later in the article series, it does not conform to any known modelling of magma chambers for stratovolcanoes.

At the same time Hekla does not in any way conform to the standard interpretation of what a fissure row is, in fact it is closer to a stratovolcano than a fissure row. A fissure row is a fissure that is creating a line of mono-genetic or poly-genetic cones erupting lava that either originates from a central volcano or originates directly from the mantle. The fissure rows emit large amounts of basalt in Iceland and form low ranges of mono-genetic spatter cones as is evidenced at for instance Lakí and Veidivötn and are parts of fissure swarms extending from central volcanoes.

In the end the defining difference is that fissure rows do not form permanent magma reservoirs negating the possibility for repeated eruptions from the same exact position. Or in other words, it will not be able to form a central volcano. And whatever Hekla is, it is a central volcano that erupts in a repeat manner.

Before we start discussing a possible new model to describe Hekla’s function we need to describe the volcano as well as we can without using any current model to get rid of our preconceptions. Imagine that we are students of volcanology that are using Hekla as our first and only volcano to study.

Taxonomy and terminology of Hekla

The beginning of the 1947-48 eruption started, the cloud reached a maximum of 30km altitude.

Hekla is a unique volcano that is the only example of its class of volcanoes on earth. It is a central volcano containing a magma reservoir of a shape that will be determined later. Above the magma reservoir an edifice is located that in shape looks like a merger of an upside-down Viking longship and a stratovolcano.

Hekla consists of a 7-kilometre-long NE/SW trending fissure called Heklugjá that is situated within a NE/SW trending fissure swarm extending 15 kilometres’ northeast and southeast of Hekla proper. Eruptions on the fissure swarm erupts distinctly different lava than what is erupted from Hekla proper. This is unique and any model would need to be able to explain this problem.

Thus we have concluded that Hekla is a central volcano fissure with stratovolcano traits. We also find that no current terminology is succinct and that a new terminology is needed to name the edifice. I hereby propose “stratofissure” as the best fitting descriptive name of the edifice.

Location of Hekla

Icelands volcano-tectonic regimen. From

Hekla is located on a complex part of the Mid Atlantic Rift (MAR) where complex tectonic forces and a mantleplume has transformed the MAR into the general Icelandic tectono-volcanic zone.

The stratofissure of Hekla is situated where the Eastern Rift Zone (ERZ) meet the Southern Icelandic Seismic Zone (SISZ) and north of where the ERZ changes into the Southern Flank Zone (SFZ).

To the northwest we find the Búrfell Tuya, an ice-age central volcano with no known Holocene eruptions. To the SSW we find the highly active Holocene volcano of Vatnafjöll, a low mountain range that has been constructed during large basalt eruptions. By extending the nomenclature from Hekla it can be best described as a “shieldfissure”.

Due to the location on the western edge of the ERZ the Hekla fissure swarm is being pulled apart roughly 4 millimetres per year. Any model must take the effects of this into account.

Magma of Hekla

The magma of Hekla is partially coming from the Icelandic mantleplume through sub-crustal flow, but it is also coming from sub-crustal melt as the ERZ is being pulled apart. The intermix ratio leans towards a larger mantleplume amount than decompression melt, as is to be expected due to the vicinity of the plume-head.

The magma is of the calc-alkali line indicating sub-crustal melting of an oceanic crust platelet that is situated below a thick volcanic over-burden. It is unclear if that platelet has been subducted or is just oceanic crust that has been covered by millions of years of Icelandic volcanism and been pushed down by the weight of it.

The magma of Hekla is gas rich with an unusually high fluorine content, during and after eruptions the ash from Hekla is detrimental to the health of both livestock and humans.

As the magma is erupted as lava it is uniquely bi-modal, or even tri-modal. The initial lava is always calk-alkali andesite, the second stage lava is semi-evolved calk-alkali basalt and during large eruptions like the 1947 eruption the third end-stage lava was simpler basalt reminiscent of mantleplume derived basalts being hotter in temperature with lower silicate content.

Any new model would need to seamlessly be able to explain the progressive shift between different lavas during an eruption.

Age of Hekla

The steaming top of Hekla. Photograph by Borkur Sigurbjörnsson. Wikimedia Commons License.

Hekla is Iceland’s youngest volcano. It is between 6 400 and 7 000 years old counted from the initial eruption. The initial eruptions did not build an edifice, instead they formed a poly-genetic row of tuff cones and lava flow fields.

It is therefore entirely possible that the edifice of Hekla is as young as 2 000 years, at least in any way we would recognize as a central volcano edifice. This is further evidenced by the rapid development of the edifice. The edifice height and shape prior to the 1947 eruption have been radically changed during the eruptions of 1947-48, 1970, 1980, 1981, 1991 and 2000.

The eruptions of Hekla

The initial eruptions of Hekla where far apart and highly explosive producing large amounts of ash and tephra fallout. Around 2 000 years ago, the rate of eruptions started to pick up with one eruption about every 100 to 200 years. Around 1 000 years ago, the rate of eruptions picked up speed and there was one eruption per 20 to 150 years and from 1947 and forward the rate of eruptions has been between 1 to 23 years.

A new model would need to explain this shift into more frequent, but less explosive eruptions.

The current type of eruptions are bi or tri-modal with an initial brief highly explosive start where andesite is the predominant erupted component. This face shifts abruptly into a second stage of evolved calk-alkali basalt, and if the eruption is large or prolonged a third stage of fluid unevolved basalt will follow as evidenced in the 1947-48 eruption.

A new model would need to explain these modal shifts.

Two problems of Hekla

There are two distinct problems that the currently used models cannot solve. The first one is the positioning of the magma reservoir. Depending on the used modelling, or the data set used you will get a new location of the proposed magma reservoir.

Pretty much every single attempt at locating the depth and size of the magma reservoir has yielded a different result. The results are not even close to each other, instead the yield results ranging from 15 kilometres’ depth up to 2 kilometres depth. Any new model would need to explain these discrepancies and be able to produce a definite answer to the shape, size, location and depth of the magma reservoir.

The second problem that must be addressed by a new model would have to be the extremely fast fractioning of the basalt into andesite. A process that is known to take considerable time in all hotspot/mantleplume volcanoes except in Hekla.


In the next part, we will see if I can hammer together the beginnings of a model that can address at least partially the problems of our current understanding of Hekla that I have outlined so far. We will also see if the new model will be testable and possible to refute.


78 thoughts on “Remodelling Hekla – A 1947 commemorative article series

  1. I am a nut for all things related to Icelandic volcanoes, especially the jewel Hekla. Kudos Carl and sooo looking forward to this series.

    Just going to drop a Hekla seismicity video quietly here… I will make a new modelling with latest data, since there were a few more quakes since the making of this video. BTW Carl if you need anything of this type of products for the series, I will be glad to assist.

      • The depth is not so shallow (22km) for intraplate earthquake. It is normally only the subduction earthquakes that are really Deep.
        What makes this into such a rare bird is that it is an intra-cratonic earthquake. This is one of the hardest and oldest spots on Earth without known fissures and faults.

    • Straight after yesterdays’ quake in South Africa. That means there is a connection. It is a very strange place for a quake. There was one M4 15 years ago in the area, but otherwise nothing significant going back 120 years! Looking at the map, I think it may still be related to the African rift further north.

      • The earthquake WSW of Pretoria is on a known old intra-plate faultline. But the one in Botswana is just “out there”.
        If it is one of the major African rifts that somehow extend down there it has skipped quite a part of Africa to be able to cause an earthquake down there.
        It may though be caused by doming of the craton, and if that is the case there may be interesting times there in a few hundred thousand years.

        • The Potchefstroom earthquake was in a known area, but quakes this size happen there decades apart. To have it followed by the strange Botswana one one day later does suggest a connection. Somehow the local stress was affected, at a distance where the first quake could have been barely felt. This is an old land but it hasn’t always been a single craton. There may still be some old stitches.

          • May be a mantleupwelling causing doming of the craton. The would explain the timing of the two earthquakes and that the known fault cracked first.
            This may be the first earthquake that has me this wound up… Intra-craton cracking caused by an upwelling in my lifetime, that would be something if it was true 🙂

        • the quakes near Pretoria are likely induced by mining, either directly or indirectly

          I think the Botswana quake is the southwesternmost extent of the East African Rift.

    • Quake definitely felt by one cousin living in Jo’burg, as a minor tremor. Not felt by another cousin living in Malawi… Which I suppose is no surprise.

  2. Thanks for the post, looking forward to the next one. A restless Helka, could the timing of this article prove prescient? 😉

  3. Some say Callaqui in Chile is an analogue of Hekla, is that true or are they different? If it is, it could help with an explanation. On another note, can you explain the relationship/interaction between Hekla and Vatnafjoll? To me, they’re independent volcanoes that seem to act like a double system…

    • Would the andesite have migrated to Hekla from a larger area, rather than having fractionated directly under Hekla? A “catchment area” 10 times as large would generate 10 times the volume of evolved magma in the same time. Flow pressure from the mantleplume would mobilise the magma and the spreading under Hekla would provide a new home for it.

      • I will come back to the rapid fractionation, but the catchment area is more related to the calk-alkali part than the fractionation.

    • Callaqui is on the Surface a lot like Hekla, but there the difference ends.
      For instance, Callaqui is a subduction melt volcano and Hekla is a MAR/Mantleplume volcano.

  4. “Any new model would need to seamlessly be able to explain the progressive shift between different lavas during an eruption.”

    There’s an important addition to that statement; it would also need to explain how it can evolve such different magmas in such short a timespan – IF that’s what it does. Hekla erupts. It starts andesitic but ends up erupting pretty much mantle plume basalt. It stops erupting. Then sometimes as little as ten years later (1) it erupts again – and woo hoo it has another batch of andesite to start the show with fireworks.

    – Does that represent the evolution of a batch of andesite from basalt in only 10 years?

    – Or does it represent multiple magma chambers? Perhaps in the first eruption it erupted only ~10% of the contents of its ‘andesite’ chamber; ten years later that chamber has recharged sufficiently for another ~10% to open the show? Does that happen?

    (1) a tie-in between repose periods and eruptive behaviour is needed too. Are first eruptions after a long repose period larger? More violent? More andesitic?

    • That might show up as inclusions in the magma. As more andesitic material becomes included in the ejected magma. If the andesitic inclusions from different eruptions are chemically related, it might go a long way in illuminating what is going on.

      We’ve seen evidence of wandering magma pockets, so it might be moving around to different reservoirs… if present. (Carl had me chase down a GPS anomaly that didn’t quite fit normal volcanic behavior several months ago.)

    • Since I do not want to pop the Cherry to soon I will only answer one question here.
      Yes, there is a difference depending on the repose time.
      The andesite volume will be smaller with a shorter repose time. The 1991 eruption came just a year after the 1990 eruption and was quite mild-mannered for being Hekla (VEI-2). And for the longer reposetimes like the 1947-48 eruptions we see a spectacular increase in the andesite phase.
      During the 1947/48 eruption Hekla coughed up 0.2km3 of andesite in about one hour creating that famous 30km towering ash column.

      I will return to the magma reservoir thing in the next part, and obviously the fractionation part.

  5. Hekla looks different from different directions. Look along the ridge and it is clearly a stratovolcano, with the typical steep sides. Look towards the ridge and it is very elongated and more shield-like (although still too steep for a shield). Basalt cannot be a major part of the eruptions, otherwise the slope wouldn’t be so steep.

    (taken from Thordasson and Larsen 2007)

    The crucial points are: the very slow but non-zero spreading rate; the location of Hekla spot on this spreading rift (normally the volcanoes are off the the side); the lack of earthquakes prior to an eruption which suggests the magma coming up from very deep (perhaps punching through the shallow magma chambers?); the eruption frequency (1 in 10 eruptions in Iceland come from Hekla); the eruption style with the large majority of magma ejected in the first hour of the eruption during the explosion, but very much less during the later effusive phase (no other Icelandic volcano does this mix). Good luck!

    • I hope that I can stitch it all together in the next part in such a manner that you do not transform me into poppercorn 🙂

    • That is weather related effects that are visible right now on the strainmeters around Hekla.

    • Hello Andrej!
      Do you Think you could do one for 1999-2000 and Another 2001 to present day?
      That would give the previous eruption and the period prior to the next one separated. 🙂

      • Yes, will do it first thing tomorrow. I will tho assimilate and add the quake data back to 1995 to increase the sample size. The quake numbers might not increase by much when I add 4 more years, since not everything was recorded back in those days (at least not lower mags), but at least the sample size for location/depth information of the pre-eruptive phase will be a bit more relevant.

        Tho it is worthy of note that the monitoring system has improved over the years, which is why we are >probably< seeing more seismic activity in the recent years being recorded at Hekla. Tho the increase is rather substantial, so its unlikely that the monitoring improvements are the main cause (if at all to a certain degree).

        • The network is 100 times more sensitive today than it was prior to the 2000 eruption. So, a significant portion is technological skew.

  6. Carl, great article and looking foward for the disclosure of your theory in part II.

    A few comments:

    1) Hekla is located in an unique spot, between a fracture zone and a large rifting region. Hengill and Theistareykjarbunga would be the most similar things in terms of their placements between fracture regions and rifting regions.

    2) Hekla is the youngest Icelandic volcano, extremely young indeed but rapidly increasing in volume until I suspect, like most volcanoes in the area, it will go caldera. Then, who knows, things will behave like all other central volcanoes in south Iceland, Vatnajokull and north Iceland.

    3) I see similarities in activity with Torfajokull, but not exactly in the same time scale. Torfajokull erupts always dual-magmas (every few centuries). In all 6 or 7 Holocene eruptions, it always follows like this: first, rhyolite magma erupted at Torfajokull proper and then large basalt floods in the nearby Veidivotn fissure, just a couple kms northeast. Yes, I know its a dike intrusion probably from Bardarbunga (similar to what we saw with Holuhraun that triggers the first rhyolite eruption and then the rifting basalts). Nevertheless, this makes me wonder whether some sort of similar mechanism could be at play in Hekla >> hotspot plume magma intrudes in the upper mantle Hekla, coming from the northeast, and triggers an andesite eruption, followed by basalts

    4) a few other Icelandic volcanic systems share some similarities to Hekla: Westman Islands and Hengill. Both appear something more than a crater row, with hints of being established central volcanoes. However none seems particularly explosive though Hengill has an interesting triple junction location.

  7. As requested, here is EQ data for Hekla for two different periods. I haven’t yet added 95-97 years, but will do so at the first chance I get.

    I have to say, its quite a difference. A lot of it comes from better monitoring system these days, but the locations/concentrating of quakes is a bit different than for the last eruption. It might still change tho, since we are still expecting an eruption.

    The biggest difference is of course the lack of earthquakes towards SE of Hekla in the post eruption period (2001-2017), but there are lots of earthquakes towards NE of Hekla along the fissure line. Funny enough, the pre-eruption period (1998-2000) had pretty much no (recorded) quakes down the fissure towards NE, where we see them now, and had the batch towards SE. So an opposite situation. What they have in common, is that the quakes go downward with depth, as you go away from the mountain, both towards NE in the latest period and towards SE in the 1998-2000 period. It will be crucial to add more data, for the years of 1995-1997 ASAP, to further confirm or disprove this difference.

    • Thanks Andrew, the lack of quakes in the SE between Hekla and Vatnafjöll is indeed very interesting…

    • And comparing cross sections again:

      (The image showing the red line of section on the right shows all quakes and is just for orientation purpose. The data shown on the left cross section is what the text says)

        • Or is it a big mushy area where solid and more liquid are in a state of constant flux?
          I know nothing… Just guessing what might make for a magma chamber that appeared to be in different locations at different times.

      • It is Always interesting when you see theoretical data being supported by real Life data 🙂

  8. 2011-Mar 2017 earthquake locations for Iceland, coloured by depth, with slight magnitude scaling.
    Shown are only deeper earthquakes (18km and deeper), which are usually associated with magma movement/transport.

  9. Good article Carl. Can’t wait for more. Just love those Icelandic volcanoes. 🙂 I think Bard tho wants attention. 2 stars

    06.04.2017 15:08:24 64.655 -17.476 0.1 km 4.1 99.0 3.0 km ENE of Bárðarbunga

    06.04.2017 15:22:20 64.670 -17.465 8.0 km 3.2 99.0 4.4 km NE of Bárðarbunga

    Info from IMO.

    Could someone put 3dBulge graph on here from Baering?

    Showing more activity with Herðubreið too.

  10. For the first time ever! Never before seen! (a bit of a dramatic intro 😀 )

    Let me present you with a time sequence of earthquakes in minutes and even seconds. It is the Hekla 2000 eruption. This video presents earthquakes data (in colour by depth) for the day of the eruption, 26th February 2000.

    What is nicely seen is the stack of earthquakes going SE and downward of Hekla, before the onset of the strong harmonic tremor and the eruption. This earthquakes could potentially show where the magma came from or where the “vent” went with depth.

    First view is from the NE looking towards SW, and the second view is normal top down view.

    I leave further interpretation to each individual. The video can be slowed down on youtube if needed. I recommend watching in HD.

  11. We have a little swarm on Mauna Loa west of the caldera. Around 18 quakes from 14:00 on the 6th till now. Depths from 0 to 7+ km, maginitudes from 1.8 – 2.6. Otherwise it has been quite on the rest of the island.


      • Actually looking at the topography IF an eruption DID start in the area of the swarm the flows would likely go the other side of Hualalai and enter the sea through the city of Kailua-Kona itself. That would be a bad spot for an eruption in human terms… but my gut tells me we’re getting closer. Mauna Loa eruptions do sometimes start with little or no warning.

        • That area is where we stay when we are there! The Sheraton where they watch the Manta’s. My wifes 20 weeks a year of traveling for work provides us with the hotel points for the stay. It just does not make up for her being gone all of that time.

          I also think I remember that these flows on the west side tend to be fast movers with the topography.


        • It should be possible to save the city with the same technique that the Icelanders used at Heimaey. Enough water and you can stop pretty much anything.
          Problem is the amount of pumps, I do not know if the US has a storage somewhere of water pumps that will survive prolonged usage with salt-water.

          • It might be doable. SW rift zone eruptions can be very fast and voluminous – reaching the sea in hours; no time to organize any response of that kind at all. But the topography around Hualalai should slow the flow a bit and give them time to get organized. They *might* be able to channel the flow into the least damaging route into the sea – but equally (being the USA) they might well get tied up in litigation from property owners who would be affected by a diverted flow….!

        • A year ago I looked at the timeline of the flow since the 1840’s for the Troube in Paradise post, it seemed to me that if the patter holds ‘Kealakekua Bay would seem next’. That would be 10 miles south of Kailua-Kona. From the location of the current swarm, downhill is almost directly at Kealakekua. But Mauna Loa rarely does what is expected.

      • Mike,
        I have some questions on what the source of these are. With the cross caldera motion stopped and the summit vertical motion looking like it has paused from its upward trend over the last several years, I am wondering if these are seismic. I do see the vents and historic lava flows emanating from the area when I look at Google earth. I did think it was worth reporting since I have not seen this kind of group on Mauna Loa since I started watching Hawaii last year. I looked at the webicorders, but I have already proven that I can’t read those things, and I have not check the weather this morning.

        We hope to vacation there again this summer, I remember us driving through those flows looking for snorkeling spots.


        • I snorkeled once. When I spotted something (a vague shape) in the distance bigger than me, I quit. This was in Hurgada on the red sea. Ya see, my philosophy on surviving volcanoes applies in all aspects of life. To avoid trouble. Don’t be there.

          Chicken? Yup. Proud of it. If I’m not equipped to deal with it, then I’m not playing.

          BTW, for the record, Hurgada has a nice reef system. Pizza at the local hotel, not so much… unless you like Calimari.

  12. According to steadfast rumours part II out of V of Hekla will be up and running tomorrow.

    It is now we will have to wait and see if Albert will hit me over the head with Karl Popper.

    • Nice one, which looks like it has send a “shockwave” of earthquakes across the SISZ. (not literally of course) 🙂

Comments are closed.