Living Dangerously – Grimsvötn Forecasted

Grimsvötn 2011 eruption. Photograph from Reuters.

In August of 2017 Albert and I stuck our necks out on a limb and made a forecast each of what the future pattern of Grimsvötn was most likely to be. Or in other words, when would it erupt next.

As I reread our two separate forecasts, I am struck by how different methodologies we employed. Both versions are deeply entrenched in historic data and science. The data employed was obviously the same, and still our different methodologies gave us slightly different results.

I ended up with the most likely time being between December 2018 and December 2019. Albert got a best fit for 2021.

We wrote our parts as a very friendly competition against each other, but perhaps our real antagonist is nature itself. Foremost it was, and is, an exercise in showing that it is indeed possible to forecast upcoming eruptions to a fair degree. Obviously, it is not possible for all volcanoes, but for those that have regular eruptions, and that are showing clear signs, it is possible.

As we revisit the volcano more than a year later, we once more pit our wits against a sea of data to wrangle out the elusive fish of a Grimsvötn eruption.

Since I haven’t read Alberts part as I write this, I will make another forecast. Albert will employ mathematics as his main tool, and I will try once more to find signs of pressure increase factors in GPS and Cumulative Seismic Moment data in the area of Grimsvötn.

Who will be right? Me? Albert? Or the pesky nature of a volcano? The game is on, but I know that both me and Albert would love to win in an ever so friendly way.

Albert, should we bet the first round of beers? Problem is just how we bill Grimsvötn if it throws us something unexpected.


Longterm trend of M0.8 earthquakes at Grimsvötn running from the 2011 runup to today. Image by the Icelandic Met Office.

There are two things that have been the same for the eruptions of 1996, 1998, 2004 and 2011. The first of those are that the pressure increase as shown in the Cumulative Seismic Moment (earthquake energy release) is surprisingly consistent, and the same goes for the GPS-data showing inflation.

This is caused by two factors. One is that Grimsvötn has the highest known sustained influx of magma into the system of any volcano on the planet. The influx ratio is staggering, between 0.5 and 0.8 cubic kilometres per year.

That is roughly 4 times more than the combined input of all Hawaiian volcanoes, and a whopping 10 times larger than the most productive supervolcanic system, Uturunku. Obviously Grimsvötn is not a supervolcano, it is not even a supererupter by the common definitions.

Still it has produced more ash and tephra than any other volcano on the planet in the last 12 000 years. It has also produced prodigious amounts of lava in the form of distal systematic eruptions in its two-pronged southern fissure swarm, around 30 cubic kilometres or so of erupted lava.

The reason behind the almost ridiculous numbers are that the highly elongated system of Grimsvötn take up more than half of the tectonic spreading in Iceland. Or in other words, the bulk of the lava is back-filling the spread at depth, what we see erupting is just the over-spill.

The amount of over-spill is though on a mammoth scale. If we combine the 150 cubic kilometres of the Saksunarvatn Tephras, The 30 cubic kilometres effused in the 3 known rifting fissure eruptions, 10 km Skaftár Fires ash, and all the other central volcano eruptions throughout 12 000 years, we end up with a conservative number of 250 cubic kilometres of total ejecta.

That equates to an average of a VEI-3 eruption every single year since the end of the last glaciation. Not bad for an over-spill, not bad at all.

The deep magma reservoir is open-ended downwards to the mantle, explaining the very small number of deep earthquakes. This deep reservoir can be visualized as the over-turned hull of a sailing boat.

The hull’s stern is situated about 20 kilometres NNE of the Grimsvötn volcano, roughly at a spot of intriguing deep earthquake activity that we have named Greip since it may be a formative central volcano.

The hull then continues for 120 kilometres SSW, continuously tapering off until it dies out around where the Skaftár Fires erupted.

The central volcanoes of Vatnajökul. Danish map in my private collection.

From this hull-shaped reservoir conduits go upwards into the central-volcanoes above, these are from north to south, Grimsvötn, Háabunga, Geirvörtur, Thordharyrna, Hágöngur and Eldgigur. In a sense of it Grimsvötn is the deep reservoir, with all these central volcanoes being sideshow Bobs.

And what glorious sideshow Bobs they are. Not counting Grimsvötn we have the serial VEI-4 volcano of Thordharyrna and the enigmatic Hágöngur responsible for the 536 VEI-6 that had a very serious impact on the climate (Katla topped it off in 540 with another big VEI-5). But that is something for a rainy day to write about.

The pertinent point is that at least Grimsvötn has open conduits from it’s deep 500 cubic kilometre reservoir up to 3 upper reservoirs in the central volcano.

It would be simple to believe that these reservoirs are situated under the 3 intersecting calderas that create the Mickey Mouse shaped caldera-system. This is though wrong, since only 2 out of the 3 calderas have a current chamber.

Map of Grimsvötn caldera with the pertinent magma reservoirs and features marked out in a rough position by me.

The North-west caldera (1) has a reservoir running up towards Gjálp (where the Gjálp lava originated), the central caldera (2) has the most active chamber responsible for the 1998, 2004 and 2011 eruptions.

The third reservoir (3) lies SSW of GFUM GPS-station in the direction of Háabunga, but it is clearly separated from the magma-chamber of that volcano as evidenced in earthquake data.

And here comes a bit of a conundrum, because the pattern has changed at Grimsvötn. And we will need to go through what that is.


The Holuhraun eruption. Photograph from Iceland Naturally.

Having several large volcanoes in close quarters has it’s effects. Especially since Bárdarbunga and Grimsvötn are the two largest volcanoes of their type on the planet.

Gjálp and Holuhraun showed the effects of this amply. In the first case a large earthquake at Bárdarbunga caused a pressure differential at the North-west magma reservoir of Grimsvötn that ripped it open in a fissure eruption.

In the second case the magma moving out of Bárdarbunga bounced against the pressure of the northern end of Grimsvötns deep magma reservoir and turned north.

Now it is analogy time. Imagine that you have two balloons. These are a bit of special balloons since both balloons are constantly being inflated until a valve brakes and a bit of air goes out. If the balloons are in different parts of a room, they will not affect each other in any meaningful way.

But if you push the two balloons hard against each other things will rapidly become interesting. In the real world there are several balloons massaging each other, and the names of the balloons are Gjálp, Grimsvötn, Háabunga and their common deep reservoir, they in turn massage (and get massaged by) Kistufell, Bárdarbunga and Hamarinn.

Longterm GPS-plot (plate-detrended) showing data from the just after the 2004 eruption, up to today. Notice the combined uplift since 2011, it is 50 centimetres. The Holuhraun effect is clear. Image by the Icelandic Met Office.

During the 1996 eruption of Gjálp the highest pressure in all of these was inside the Gjálp reservoir of Grimsvötn. As the relatively low-pressure volcano of Bárdarbunga suffered that large non-double-couple non-volumetric-change earthquake it lowered the pressure in Bárdarbunga creating the possibility for the Gjálp reservoir to rapidly expand, this in turn opened up the fissure nearby and general squirting ensued.

After 2011 all the reservoirs at Grimsvötn was relatively low pressure, but Bárdarbunga was medium-pressure and an oddball turned out to be high pressured. As magma pushed up from depth into Kistufell that volcano burst sideways towards Bárdarbunga, creating an extra-volcanical dyke running straight into Bárdarbunga insta-flipping it from a medium-pressure volcano into a high-pressure system, Bárdarbunga then cracked, and a new dyke ran off towards Greip at the northern end of the Grimsvötn deep reservoir, there it either connected, or rebounded, and went towards Holuhraun.

Not only did this magmatic pinball machine create a spectacular eruption, it also had grave consequences for the timing of the next Grimsvötn eruption.

As the balloons of Kistufell and Bárdarbunga erupted they lost pressure and shrunk, quite a bit. As they did that, they squeezed Grimsvötns magma chambers less and less. This meant that the Grimsvötn reservoirs could expand in the direction of the two deflating central volcanoes ever so slightly. This expansion lowered the systemic pressures of the magma chambers at Grimsvötn.

This stopped the build-up towards the next Grimsvötn-eruption dead in its tracks, or at least created a 2 year long hiatus in pressure build up as Grimsvötn took up half of the work of refilling the 2 cubic kilometres erupted out of Holuhraun from a pressure standpoint.

On this plate-detrended GPS-plot we can clearly see the point where the pressure returned to full and rapid inflation sets in. That puts the point of no return for the upcoming eruption to July. Image by the Icelandic Met Office.

99.9 percent of all volcanoes on the planet would by now have been out of commission for a century or more while doing the refilling, for Grimsvötn it took less than a volcanic heartbeat since it’s the grandmaster of magmatic influx.

In July the pressure reshuffling was finished as pressure had gone up in both Grimsvötn and Bárdarbunga to create a return to the normal hug-and-squeeze-fest between them.

But before we go there we have to talk about another profound effect that Holuhraun had on Grimsvötn. Prior to the 2014 eruption it was the central magma chamber that was inflating in the normal fashion, and we where heading for an almost certain repeat of 1998, 2004 and 2011.

But as the pressure fell in that chamber something unexpected occurred. The Nunatak upon which GFUM GPS is situated started to go up and being pushed to the north-east. This indicated strongly that it was the magma reservoir directly south-west of the station that was inflating. This was also evidenced by earthquake data.

Now it is time to see what has happened in the last half year, and what the effects will be upon my August 2017 forecast.

Grimsvötn revisited

We can safely say that pressure was fully restored in the systems as of early July 2018, at that moment a period of uplift started that so far has raised GFUM 10 centimetres in less than half a year. Total northward motion is 3.5 centimetres, and the maximum east motion was 3 centimetres.

If we just look at those numbers, it would be easy to assume that the next eruption would be south-west of the station. But it is not as easy as that.

Because a month ago the central upper magma chamber caught up and started to affect things quite a bit. The northward trajectory stopped dead in its tracks, and eastward motion reversed completely, and the poor GPS moved a centimetre west in under a month.

Currently, it is impossible to say which of these two magma chambers will take care of the eruption. It does though seem like the central chamber is the stronger of the two and should therefore be most likely to erupt.

On the other hand, those two could be connected and the eruption could utilize the magma from both for the upcoming eruption.

If the eruption would emanate from the southern and less well-known chamber there is a slight possibility that the chamber roof could fail if the ensuing eruption is large enough, and Grimsvötn is quite good at landscaping at times. One should though remember that it is a highly uncommon thing.

Cumulative Seismic Moment

The all important CSM-plote. Note how the earthquake energy release started to go up steeply as the pressure grew to large. Image by the Icelandic Met Office.

As the pressure increased between the systems the inflation and pressure started to increase rapidly, and it came as no surprise that the amount of energy released by earthquakes started to increase rapidly.

From having had the flattest trajectory ever recorded, the CSM-plot roared back with a vengeance as can be seen on the image above.

If the new trajectory would hold true, we could indeed have that December 2018 eruption, but it is more probably that it will return to a trajectory closer to 2004, and that would mean that an eruption would occur somewhere between April and July 2019. If the trajectory calms down into the 2011 version, we would see an eruption at the end of 2019.

I am here using the assumption that the necessary CSM-value should be between 3 and 4.5, that might be a bit off, but I do not think by that much since we have never seen anything pointing towards that.

Obviously, the trajectory could turn into something flatter than I envisioned above, if so the eruption would come at a later date more along what Albert originally deduced mathematically. And, obviously there could come a large prolonged swarm taking care of things within days or weeks, if so, it would cut the time short with a vengeance.


In the end I find that I do not need to change my original forecast that much given the data at hand. I think that December 2018 might be a bit optimistic, but at the going rate anywhere between April and December 2019 seems to be a good bet for a bit of volcanic tourism for those of the ashy persuasion.

Next week we will have Alberts version. And I for one am highly curious where the mathematical approach will lead. Upwards and erupwards I am sure!



209 thoughts on “Living Dangerously – Grimsvötn Forecasted

  1. n earthquake M3.5 occurred at 21:37 (23 November) approximately 10 km E of Hamarinn in Vatnajökull glacier.

    • 7.4KM depth too 🙂
      Source: IMO
      Friday 23.11.2018 21:37:22 64.500 -17.598 7.4 km 3.5 99.0 10.2 km E of Hamarin

      • “An earthquake M3.5 occurred at 21:37 (23 November) approximately 10 km E of Hamarinn in Vatnajökull glacier. No volcanic tremor has been detected. Earthquakes of this magnitude are occasionally recorded in the area.”
        Written by a specialist at 23 Nov 22:48 GMT

  2. The Bardarbunga rift and dyke and sill system extends 200 kilometers from the central ice clad volcano itself.
    In the 1400 s Bardarbunga sent a massive Holhuraun like intrusion south .. hot basaltic magma intruded into Torfajökulls stale ryholite magma chamber and pushed out that old magma.
    Bardarbunga force feed Torfajökull in 1477. That was a pretty long dyke! all way down to Torfa.
    Many km3 of intruded and did not erupt.
    Had bardarbungas 2014 dyke been even more vigorous it woud maybe cut into the Askja magma system.
    Im supprised that Holuhraun was so small ( 1,6km3 ) Bardarbunga is known for very large infrequent rift flood lavas 6 to 30 km3.

    • 1477 was pretty big for an eruption where it was. Most eruptions on veidivotn are smaller and more like trollargigar in the 1860s, holuhraun was small compared to 1477 and thorsja but it was pretty big. Bardarbunga is not grimsvotn no#2 it is much less active in the long run and it doesnt average 10+ km flows more like 1 km3 flows with one big outlier.

    • Practically clear blue skies at Mauna Loa right now, rain is clearly not to blame this time:

    • Looks like the strongest signals are on the recorders from 8 – 12, 000 feet ASL around the summit.

  3. I Loves Icelands landscapes!
    These dark, moody, barren craggy hily landscapes with their gloomy skies are amazing.
    Iceland reminds me of Tolkiens Middle Earth or Warcraft III. Iceland is a dark fantasy come true.
    Icelands landscapes are like taken from a saga. Its hard to visit Iceland without thinking of orc and goblins that patrols the mossy landscapes and dragons that lurk in the mountains.
    Iceland is perhaps the worlds most beautyful country. Its defentivly the photographers favorite.

  4. Some landscapes in Iceland are totaly bissare!
    The Skaftaeldahraun 1783 lava flows are one of the strangest.
    The strangest landscape I ever seen. These piles of mossy lava rocks.
    Its almost like the landscape was made of moldy bread.


      Here is a fun video for you.
      And the only video I ever seen of that eruption. Almost 60 years ago.
      This basaltic eruption came from a ring fault around the inner caldera.
      My mum wasborn that year. And 35 years after that I was born in 1995.
      The ponded pahoehoe flows in this video is very fun! with overturning crusts and lakes
      Socuh things are often seen in ponded hawaii flows

  5. Grimsvotn 2011 erupted 1 km3 of magma in a week, that would give it an average eruption rate of 140 million m3 per day, or 280 times what kilauea averaged during its eruptions from pu’u o’o, and 10 times higher than fissure 8 or holuhraun.

    I think most of the magma was erupted in the first day though, which would give it an eruption rate of about 800 million m3 per day.
    That gives it an eruption rate of 9259 m3 per second. In going with my rough idea of a hawaiian-type lava fountain being about 1000 m3/s to reach 500 meters, grimsvotn would have theoretically done a fountain that was 4600 meters high… It doesnt work so well for more explosive eruptions but the fact still remains, 2011 was like no other basaltic eruption in the 21st century. This is what the start of really big basaltic eruptions actually looks like, skaftar fires would have been like the 2011 eruption but after the plinian activity there would have been lava.

    Full sized flood basalts like the roza flow eruptions would have been far from effusive in their initial stages, even if only 1% of the eruption is explosive that is still a large VEI 6 eruption twice as big as pinatubo, and based on the ratio of similar eruptions in iceland in historical time, 1% is probably on the low side, so large flood basalts probably started as small VEI 7s with all the associated features before the real show began. Basically a flood basalt is the entire spectrum of basaltic volcanism in one eruption happening at the same time and being scaled up to the point where everything else looks like an accident…

    I have also considered that maybe a number of really big volcanoes on other planets are actually centralised flood basalts, as the crust of mars is stationary (venus technically is too but might have some tectonics and probably does have a moving crust) and as such a big plume might lead to a supersized shield volcano rather than a rifting plateau like on earth. This is analogous to hawaii but with no plate movement to remove the volcanoes from the source. If hawaii was under a craton like africa which doesnt move much then it likely would have created a monster shield volcano, the tibesi province in the middle of the sahara is similar to this, and has been compared to elysium plateau before, and has some really big volcanoes, but hawaii is much bigger still, and is far more productive over geological time. The volume of lava erupted by the hawaii hotspot in the cenozoic is actually similar to olympus mons, and about half of that is less than 3 million years old, so if the hotspot was under crust that was suitably thick and instead of breaking it up it melted a hole, there could be an absolutely monumental volcano created on earth. This could have actually been exactly what the deccan traps was, the eruption rates I calculated for it are way too hig hfor it to seem feasible for the eruption to actually stop at any point, so instead of erupting at 5 digit numbers m3/s for short times, it erupts at 1000-2000 m3/s containuously and builds an enormous lava field that logically would have been somewhat centralised. The biggest flows could have been when ponded lava near the vents was freed by eruptive fissures rupturing the walls of the pond and allowing thousands of km3 of lava to flood out. Would have been a terrifying sight for any dinosaurs around at the time that were still alive after all that SO2, and this 10-20,000 year period of highest activity does include the K/Pg boundary unsurprisingly.

  6. Grimsvötn 2011 was so powerful the first day and so quickly that it prouduced huge lumps of very mafic Thoelite basalt pumice. There are huge thick black pumice beds near Grimsvötn from that event.
    Mafic pumice are are as most often the gases bubbles out the very fluid lavas.
    2011 was erupted so rapidly that gases coud not escape and bam!
    herefore means that the magma rised very quickly from deep beneath the volcano. This is consistent with the very intense eruption. Eruptive rates the first hours was more than 11 000 cubic meters a second.
    Lots of lapilli and and ash was produced too.
    2011 was a very impressive event and Grimsvötn can do much more.
    There been many souch Grimsvötn eruptions in the past some are good sized VEI 5

  7. Sviahnukur ( Grimsvötns south caldera wall that pokes out the icesheet )
    Likley one of the places in Iceland that haves the very badest weather.
    I imagines never howling winds and biting cold and severe frosts.
    Not a place where you wants to have your house I think.
    Brrrrr frostbite ( I guess that place haves an Icecap climate )

Leave a Reply