Grimsvötn – The Saksunarvatn Tephras

A repost on the past of Grimsvotn, as it seems to be nearing its next eruption, albeit not on the same scale as the one described here!

Aerial view taken 21 May 2011 shows the eruption of the volcano Grimsvotn in the south-east of Iceland. Do note the majestic shockwaves visible in the clouds. EPA/EGILL ADALSTEINSSON

Aerial view taken 21 May 2011 shows the eruption of the volcano Grimsvotn in the south-east of Iceland.
Do note the majestic shockwaves visible in the clouds. EPA/EGILL ADALSTEINSSON

It is sometimes hard to understand the size of the Icelandic volcanic systems. We often read statistical things like “Half of all the ash in Europe” and “One third of all basalt produced in the world” and we still do not really get it because we lack a point of reference.

Instead we time after another get stuck with what looks big on the surface, such as large stratovolcanos in the Andes or big caldera formations like Yellowstone. But, by looking at what we see we miss the big picture.

Let me start all over again at the beginning. What is a volcano? A volcano is in one aspect like an iceberg. Most of it is hidden. When we think about volcanoes we have to start at the beginning of them, and that is where the magma comes from that ultimately forms the small part that we can see.

In Iceland there are two processes that produce magma. One is basalt formed as the MAR (Mid Atlantic Rift) is spreading. That is called MORB-formation (Mid Ocean Ridge Basalt). It is basically decompression melt that occurs as the plates are pulled apart at the MAR. This is what creates most of the magma on the Reykjanes Peninsula and north of Theistareykjarbunga.

The other process is the nascent mantleplume that is burrowing ever downwards into the mantle. How it started is not well known, all we know from petrochemical evidence is that it over time has increased in flow-mass and depth. And that it is fairly stationary at the center of spread in Iceland.

Basalt formed from the mantleplume is hotter and petrochemicaly different compared to the MORB derived magma. This kind of basalt is typically found in relatively pure form in the Bárdarbunga and Grimsvötn volcanic systems. There is a general trending away from those two towards the outer edges of Iceland where gradually MORB increases and plume derived basalt decreases.

The deep reservoir

Grimsvötn volcanic system consists of 6 en echelon central volcanoes and a fissure swarm that is between 150 and 190 kilometers in length. Here the fissure swarm is imperative for understanding the deep magma reservoir. As Iceland is pulling apart with 0.9 centimeter per year at Grimsvötn an upside down canyon is formed from the mantle and up under the volcanic system. As it spreads the canyon is filled with magma that enters the system mainly via the mantleplume, but partially from MORB-processes.

This process explains how a 400 cubic kilometer deep reservoir can form under the Grimsvötn volcanic system. The same would obviously also be true for the Bárdarbunga system. So, under an area of 190 by 50 kilometers we have roughly 800 cubic kilometers of magma.

The magma reservoirs

Plate tectonics in Iceland. Wikimedia commons. Note that the Eurasian plate is relatively stationary compared to the North American Plate. The Micro-plate mentioned in this article is delineated from Reykjanes Peninsula, up to Langjökull, via Hofsjökull and into Bárdarbunga/Grimsvötn where it then turns to the SW and goes through Myrdalsjökull and onwards past the Vestmannaeyjar.

Plate tectonics in Iceland. Wikimedia commons. Note that the Eurasian plate is relatively stationary compared to the North American Plate. The Micro-plate mentioned in this article is delineated from Reykjanes Peninsula, up to Langjökull, via Hofsjökull and into Bárdarbunga/Grimsvötn where it then turns to the SW and goes through Myrdalsjökull and onwards past the Vestmannaeyjar.

From the deep reservoir conduits lead up into shallow magma reservoirs that are commonly called magma chambers. In the case of Grimsvötn the high rate of eruptions has led to these being constantly open, so there are few and small earthquakes occurring as magma move upwards. We also know that there is more than one shallow magma reservoir under Grimsvötn from GPS-trajectories during eruptions. In other words, Grimsvötn often alternates between different magma chambers during eruptions.

Grimsvötn proper

Now we have arrived at what we see as Grimsvötn. It is a complex volcano that has little resemblance to what we would think of as a strato-volcano. In fact it is a strato-volcano with Tuya-formations, shield-formations and dome-formations all at the same time. But, it is also a caldera-formation. Currently we know of 3 separate calderas at Grimsvötn and that is the reason for the peculiar form of the caldera-system.

As the last ice age wound down the isostatic pressure relief caused a marked uptick of eruptions in Iceland. The pressure from the ice-cap diminished the influx of magma and few eruptions occurred. This caused magma to pool for a longer time in the deep reservoirs, but also in the shallow magma chambers under the central volcanoes.

As the ice melted and the ice-age ended this caused rapid formation of new magma at depth and hot fresh magma to intermingle with old stale crystallized magma. It also caused lowered pressure in the shallow magma chambers. In a few thousand years some of the largest effusive eruptions occurred on the planet during all of Holocene. It was for a time even believed that one or more VEI-7 eruptions happened in Iceland.

The last part has though been hotly debated. The opponents point to the crust being too thin to be able to withstand a large enough shallow magma reservoir, and the proponents shouted back that the amount of ash released in some eruptions was so big that they had to come from a VEI-7 eruption.

In this fight the biggest stick thrown around was the Saksunarvatn Tephra with a face value of 150 cubic kilometers (DRE)*.

The Saksunarvatn Tephra

First of all, the Saksunarvatn is not to be found in Iceland. Instead it is a lake on Streymöy in the Faroe Islands. The tephra was not found by geologists, instead it was found in 1968 by Waagstein and Johansen.

The ash layer has since been found in bogs and lakes all over Scandinavia, Denmark, Germany and in ice-cores on Greenland and Spitsbergen. It is nowadays used as an important stratigraphic layer when dating eruptions, archaeological sites, paleobiological sites and for climate research.

The problem is that there has been a long standing state of confusion caused by the interdisciplinary nature of the find. Basically it has been biologists and archaeologists that have studied the Saksunarvatn layer, and they have just said “look at the pretty sand, it looks like one big layer”.

It took all the way until this decade before a few volcanologists took a look and said “wait a minute, something is funky here”. And then some of the big guns of Icelandic volcanology were on it like bees on honey.

Now, go and get something to drink and sit down and relax. We are only going to rewrite volcanic history and partially volcanology itself. So, no big thing at all.

The Saksunarvatn Tephras

Saksunarvatn in the Faroe Islands photographed by Brian Aslak.

Saksunarvatn in the Faroe Islands photographed by Brian Aslak.

I read a couple of hundred scientific papers a year in the field. The vast bulk of them are produced so that various researchers can aid their careers to plod onwards to Professor-hood. They tend to be unimaginative, technical and most often debates ad nauseam some obscure detail. Then there are those papers that actually move science a bit forward and make me happy to read.

Once upon a blue moon I get across a paper that changes science, the ones that leaves you sitting in cold-sweat staring blankly in the wall uttering intelligent things like “duh” while longing for a good shot of whiskey since you have to remodel all that you thought you knew.

“The evolution and storage of primitive melts in the Eastern Volcanic Zone of Iceland: the 10 ka Grímsvötn tephra series (i.e. the Saksunarvatn ash)” by David A. Neave, John Maclennan, Thorvaldur Thordarson & Margaret E. Hartley is one of those very rare birds.

Be warned, it is not an easy read. It is a highly technical paper on petrochemical analysis. Instead of going into the technical details of this very well written paper I will instead move unto the consequences in an understandable way for the layman.

Let me first start with this, the Saksunarvatn Tephras where laid down in 6 large eruptions ranging from 1 cubic kilometer to 30 cubic kilometers. The figures here are the most conservative estimates. One of these five eruptions travelled in the direction of the Faroe Islands and the other four travelled towards the northwest.

Now it becomes even more interesting. The timeframe for these five large eruptions is 500 years. That means that in a very short geological time-span Grimsvötn suffered two VEI-5 and 4 VEI-6 eruptions with a combined conservative output of 150 cubic kilometers Dense Rock Equivalent (DRE).* This is a rate of large eruptions unheard of in volcanology.

There is more. The petrochemical analysis shows that the magmas formed Chrystal inclusions at very different depths and temperatures. The scientists that wrote the paper very markedly point out that a partial explanation is likely to be found in our models of heat and depth for formation of some of the samples, but this is just a call for more scientific studies in their field. These ambiguities are though comparatively small and we end up with crystals that have formed in different temperature melts at varying depths.

This leaves us with a magma composition that formed inclusions between 1140 to 1300 degrees and at pressures warying from 1 to 7.5kbar. Or in other words, some of the magma was old mush filled with melt inclusions and that had undergone evolvement, and other magmas was hot unevolved magma with less melt inclusions. In even simpler terms, during this set of eruptions magma from 2 to 15 kilometers in all stages from stale to fresh plume origin squirted out intermingled with each other.

This more or less ripped out the shallow chambers and the conduits down to the deep magma reservoir. In turn the deep magma reservoir answered by pushing up magma on an unprecedented scale. Even though the magma production in that part of Iceland is very large it does not explain what happened and the caldera formations are too small to explain it.

One explanation would be that a Graben would have formed all along the Grimsvötn Volcanic System with an average depth of 39 meters, but there is no evidence to be found now for that.

The other possible explanation is that locally plate tectonics switched direction for a while. Obviously I am not talking about this comparatively small eruption pulling Eurasia and North America towards each other. Instead I am talking about the micro-plate that exists in Iceland moving eastwards for a while.

It is likely that both things happened at the same time. By now the depression would have either uplifted or been filled in by Iceland’s large rifting fissure eruptions (Laki for instance). In regards of the micro-plate moving afterwards hypothesis, I can only say that for a while afterwards there was a few rather large eruptions at the western margin of the micro-plate. Largest among those is the Skjaldbreidur shield volcano that started to form 500 years later.

The Aftermath

A small part of the Skaftár Fires fissure (Laki).

A small part of the Skaftár Fires fissure (Laki).

After this eruptive sequence Grimsvötn was left a shadow of itself. The volcanic system was gutted and 37.5 percent of all the magma was spent. Any normal volcano with a normal rate of magmatic influx would now have gone into dormancy lasting tens of thousands of years.

If Grimsvötn erupted in the following thousands of years is unknown, but the first known return eruption happened 3 800 years later (4 550 BC) and it was a big one. It was the 9.4 cubic kilometer Botnahraun that formed the Lakí Mountain among other things.

After that Thordarhyrna had two large effusive basalts in the same size range, namely the Bergvatnsahraun (3 550BC) and the Raudholar/Brunuholar eruption of 1 950BC.

After that small explosive eruptions started at the volcanic center ranging up to VEI-2. This 3 800 long period of small scale volcanism ended with the 15 cubic kilometer Skaftár Fires at the Lakí Fissure.

During those 10 000 years the volcano had not only replenished all of the lost magma during the Saksunarvatn Tephras, it had also replenished the large effusive hraun-eruptions. After Lakí it was supposed that larger eruptions were unlikely, but once again Grimsvötn had a surprise up its sleeve.

In 1996 Grimsvötn started a series of VEI-3 eruptions ending with the even larger VEI-4 eruption of 2011. This means that the system is now fully reformed and this has consequences about how we look at it in the future and this is something that I will get back to in the next part.


*At Volcanocafé we have the good fortune of having Professor Albert Zijlstra as esteemed administrator and contributor. Albert pointed out when he saw the article prior to publication that there seemed to be an error in the figure of 150km3 DRE output in the eruptions.

He came up with very strong arguments, but in the end I used the figures given in the papers up above. I was though intrigued and doggedly dug onwards and in the end I had made Isopac-projections and calculated the volume of the 3 calderas at Grimsvötn. The result ended up between 60 and 120 cubic kilometers DRE, a figure I at least am happy with.

Thank you Albert for forcing me to not be lazy while writing and making me do the calculations myself, in the end the article is much better with this small addendum.

Another point I wish to do, the 500 year timeframe is probably a bit over the top, the ice-core samples seem to indicate a shorter timeframe.


415 thoughts on “Grimsvötn – The Saksunarvatn Tephras

  1. Just came across this – excellent photo from the Iceland Civil Protection helicopter showing the latest fissure eruption – ?v=7

    • The 3 arrows from left to right are Blue Lagoon, cut off of the hot water pipeline and finally the road.

  2. There has been an update in Icelandic today from IMO. This is the key paragraph.

    “Expansion continues under Svartsengi and model calculations based on GPS data from 3.-6. March shows that about 1.2 million cubic meters of magma have been added to the magma chamber these days. So, in total, more than 10 million cubic meters of magma have accumulated in the magma chamber. The situation is therefore similar as it was before the magma run on March 2.”

    • If the system still is open as in January and February, the breakthrough should happen with the same volume. The physical resistance of rock above the sill hasn’t changed. The eruption happens when the Newtons (or Pascal) of the intrusion overcomes the Newtons of resistance.

      • Except that the ductility may change. It is noticeable that the inflation is increasing at the point of each eruption, although the IMO model is still suggesting a similar volume as expected. There is greater error though, as previously the model also included INSAR data accordingvto IMO, but weather conditions are preventing that I would imagine.

  3. This is a plot of the Tephra sites used in Tephrabase, a storehouse of geochemical tephra data obtained via multiple interdisciplinary researchers.

    If you wish to use any of their data sets, the citation for each is provided when the data is queried as to source and research group.

  4. Quake charts are blank. Might be soon North of Grindavik. I’ve just that feeling, which the magma doesn’t find key in its decision making.

    • We should see quakes if it pops off above or in Grindavik, less so if it’s in the already warm area between Stora-Skogfell and Hagafell.

    • IMO writes in the update: “An eruption could occur with very short notice, possibly less than 30 minutes.”

      Does the continuous heating of the surrounding rock reduce seismicity in the area? This would mean that an eruption can happen relatively suddenly without a lot of seismic noise.

      • More so that rock will still need to be broken in and around Grindavik in the uppermost crust, whereas where the rifts have already appeared along the dykes should be in theory still malleable and require less energy to break through. They won’t fully solidify for at least another 6 months.

        • Does the heat make the rock more soft? I’d expect that rock contains some materials that change their behaviour with rising and enduring heat.

  5. Both Eldvorp amd Svartsengi statiins have gone above what nearly erupted the other day. The fact everything is so quiet gives me the impression when it breaks there isnt going to be much warning at all.

    If it erupts where the last intrusion went then we are probably looking at an ocean entry. The last eruption would have reached the ocean had it been located to flow that way. And the way south is both steeper and now impounded by the Grindavik barrier, if lava reaches the ocean within an hour is entirely likely.

    • How fast can lava run during the first hours? Is it comparable to normal human adult sprint speed of 20km/h?

      • Depends on the terrain, but from a quick google search typically not faster than 10km/h

    • The average wind speed at Grindavik has stayed above 10m/s since March 5, with gusts up over 20m/s. If you go up to Hagafell, average wind has been above 15m/s with gusts up to 35m/s. Looking at skjalftalisa it seems like the drop in quake activity correlates perfectly with the increase in the observed wind speeds.

      The fact that it’s quiet probably just means that the microquakes are drowned by the wind.

      • Not sure if the Husafell cam is still going but that might be the perfect view for this next eruption as the Thorbjorn cam isn’t really high enough up to look over the side of the mountain and also pretty close to the west side of Hagafell. Langhryigur cam could be a decent one to watch also. That is if it occurs along the east side of Hagafell as expected.

      • I was also checking forecast on Windy dot com , rhe wind won’t start dying down until sometime Saturday

  6. Carl, thank you for the very informative article. I’ve only been reading here for maybe a couple of years, so I didn’t know any of that.

    The use of the word “petrochemical” kind of jumped out at me. I don’t understand something there.

    • My guess (as a complete ignoramus) is that, rather than referring to products derived from petroleum or natural gas, we are seeing a more literal meaning of the word; petro (from a greek word meaning stone) and chemical as we understand it. The chemical make up of stone, even though the stone in question is liquid! Similarly, petroleum is oil from stone,

  7. M2.81 just above Grindavik. New breakout point for the magma?

    • It’s getting close now. Let’s see if it breaks surface this time and keep our fingers crossed it stays away from infrastructure and homes.

      • Before the last actual eruption there were quite a few quakes close to that road 43. I was half expecting the eruption to be close to that road at that time but in the end it chose Sundnuker area. Someone said a day or so they thought that eventually they expected an eruption in the Eldvoep area. I expect it will be sometime before that happens but I also see it as a possibility in the furure. Many of the quakes last time, and also lately, have seemed to be trending westward. Not so many as in the dyke area of course.

        • I was going to have an early night but will give it another hour now. Not really expecting anything exciting to happen but it cannot be far away as even Eldvorp GPS is almost back to height again and others are exceding previous heights.

    • TIME 20:29:39 2024-03-08
      MAGNITUDE 2.81
      DEPTH 5.225 km
      AREA: Suður­land – Reykja­nes­skagi – Svartsengi
      NEAREST Volc.system:: 8.3 km. WSW from Fagradalsfjalli
      Town:: 1.6 km. N from Grindavík (3535 pop.)
      Seismometer:: Grindavik (1.4 km)
      GPS Station:: Grindavík (0.4 km)
      IMO ID:: 1330399

      Well, this is interesting. According to the IMO we are now measuring quakes to +/- 1 meter depth.

      Since the depth is 5,225 meters deep, this is at the sill emplacement depth, so perhaps that magma is seeking to start a new dike? Time will now tell, but magma at 5,225 meters deep is no where near the surface.

      • The uncertainty in the depth is so large that you probably can’t trust even the first decimal. What we can say is that the initial break of this quake happened at approximately 5km depth. Yes, you get more decimals in the calculations, but they’re not useful. You can increase the precision using double difference relocation methods, but even then I doubt you can get down to single meters in precision.

        • and even if you could, would it actually change anything?

        • The earthquake does not occur at one point only. It ruptures an area over a range of depth and length. The larger the quake the larger the area, so the larger the depth range over which it occurs. Small quakes may seem better, but the accuracy of the depth gets less for smaller quakes as the signal is weaker. It would only make sense to state a depth to one meter for a quake that is so small it could not be measured. So best to only use depth of the stronger earthquakes, and typically a depth accurate to 1 km is fine.

          • Albert, what has been stated above is basically true. The issue apparently then is with the IMO data, is it not?

            When I stated “this is interesting”, I honestly do NOT know what to make of this, since it is a rather new development, in the listing of quake depths. I will assume that the hypocenter is using the centroid average, as a quake is not a spot point event. I know that the equipment and software techniques are getting better (probably using AI technology) so that the depth measurements are much better than in the past.

          • But IMO hasn’t changed anything. The thing that changed was that you found a new place to get the quake info and there the rounding was to three decimals. I warned before I pointed you in that direction that the uncertainty is too large for the decimals to be relevant.

            The hypocenter and centroid are two different things. The hypocenter is the point of initial break, based on the first waves that reach the seismometers. The centroid is sort of the average location of the quake, like a center of mass.

          • Albert, while that’s true, for small events – smaller than ~M5 – they can essentially be treated as point sources, and the waveforms reflect this.

            For larger events, some can still behave and be analysed as point sources; others have clear evidence of a complex rupture, and that is very clear in the waveforms. The ultimate ‘complex event’ is Kaikoura, and I have analysed it several times.

            But you’re entirely correct, it IS very difficult to constrain the depths of shallow events – and that’s largely due to network density issues, plus velocity model uncertainty.

          • It is your job, so you know best exactly what is achievable! To get a distance accurate to 1 meter would require an arrival time accuracy of order a msec (assuming the wave velocity is constant). Say a sampling rate of a kHz which the instruments probably can do. But any noise would confuse the determination, which is why this requires strong signals or many individual seismometers. For a shallow event, the biggest problem may be in converting the measured distance(s) to separate horizontal and vertical distance? For a shallow and weak event, getting enough seismometers to see the signal may be problematic. From the earthquake parameters published by IMO, my impression was that the accuracy of the depth was around 100 meters for a M3 earthquake. For much weaker earthquakes, it seemed more like a kilometer. If there is an earthquake swarm, everything gets much more confused. But that is my expectation: you know the real situation much better than I do.

          • Albert, 1KHz is a commonly achievable sampling rate in seismic digitisers.

            But the sampling rate and clock accuracy aren’t the determining factors for shallow EQs; the entire problem is:

            1. the difference in the raypath between – say – 2km and 3km deep – is miniscule, greater than the approximation in measurement caused by:

            2. The inherent uncertainties and approximations in the crustal velocity models used.

            Those are the limiting factors here. Whether the quake is small or larger doesn’t affect those; larger quakes with more impulsive, less emergent arrivals are easier to *pick* – but no less uncertain about the *validity* of those picks.

  8. As my old grandma use to say: “a watched volcano never erupts.”

  9. Something I am wondering about. It has been awhile since the IMO released the inSAR map of deflation/inflation changes of the surface terrain, also any general map showing the most recent inflation changes near Svartsengi. I hope that we get updated soon.

    • Campi Flegrei had some inflation, as I’ve read in a German blog. According to that a ship ran aground, because the uplift of the ground makes seawater in some areas more shallow.

  10. Seems the GPS has now gone beyond the March 2nd intrusion without anything happening. Its way too early het to consider if the sill will rupture somewhere else and make a new rift but if nothing happens by next weekend it could be a lot more likely.

    Problem is that if the above situation doesnt happen then the only real option is a much larger eruption. Theres not much room left in the rift underground at Sundhnjukur, not enough to contain much more than what happened last week.

    Eldvorp covers about 25 km2 of land in lava. Illahraun and Arnarseturshraun combined cover about the same, and together they cover a bit under 50 km2 of land. The total volume is likely over 0.5 km3 and that is separate from the possibly comparable volume erupted at Eldey and Stampar. So far we are at around 0.05 km3 at most and about 8 km2 covered, basically we are maybe only like 10% of the way through this and that isnt counting Fagradalsfjall probably still being active and both Krysuvik and Eldey/Stampar being intrusive and likely to erupt within the decade too…

    • I guess it is theoretically possible that Mt Þorbjörn is a relatively effective stopper for the Svartsengi / Þorbjörn system? When magma cannot rise, it meanders over to the intrusion / graben. If correct, is there a limit to how long Þorbjörn can act as a stopper?

      • I imagine Thorbjorn and the other hills there are probably off limits. I know Fagradalsfjall erupted under a hill originally but for the sill to drain at Sundhnjukagigar it needs to go under the hills so if they were weak spots we would know.

      • Thorbjorn has deep roots, I think it blocks the natural pathway for magma to flow, so it usually finds a way through just north of it. I don’t think magma is collecting underneath however, the centre of the sill (going off the GPS data) appears further west.

        • Its a hyaloclastite mountain though, so formed in the last glacial and probably after the last maximum given its preservation. So is basically gravel sitting on rock under it. There will be a degree of burial around the edges from Holocene lava but not much in a relative sense. Its just because a hill will weigh down on the terrain more so unless the hill is a polygenetic volcano it will tend to be avoided by magma. The sill complex is 5 km deep or so which is like 20x or more than the height of Þorbjorn so likely oblivious to this affect but the dikes are much shallower at the top so feel the terrain.

          Its also probably why there are so many tuyas in Iceland that all seem to have just decided to stop at the same time, it isnt because they are actually dead its just that erupting through the mountain is harder than a fissure adjacent erupting instead, and most if them are probably just vents of familiar systems that are pre-Holocene. If Herðuberið does erupt it probably will be a fissure on the base somewhere. At least assuming the seismicity there isnt just induced by Askja and tectonic, which is more likely.

    • Do we watch the shifting of the eruption from the opening phase (1st to 3rd eruption) to a second phase? We have to take into account that the eruptions series changes behaviour from time to time. The physical conditions may change a bit with significant effects on the “volcanic theatre”.

      The eruption has already left the path of app. 25 days recurrence time. This indicates a possible change and new pattern.

      • Alternatively, the event last Saturday was the next part of the same pattern and it is reducing by two days each time. If so, then maybe another 13 days to the next event. Of course, we may also be attributing a pattern that isn’t there. After all, as humans, we like to look for patterns to explain things.

        • A few observations, and rough interpretations, on this subject; as time goes on the more the tectonic stress (dilation from the plate rifting) in the Grindavik dyke(s) area (GD from now) is relieved due to the infill from every intrusion of magma. As this stress field relaxes, as long as the sill inflation is constant (this seems to be roughly true at the moment), then other areas further away from the Sundhnúkur area may become more viable for an eruption.

          Without knowing all the values of the various stress fields along the peninsula, it’s hard to know if we are at the point of a ‘jump’ or not. The centuries of built-up rifting stress was obviously high in the GD area before November 2023 but, looking at the GPS ‘stepping’ (jumps in E/W, N/S directions) with every intrusion, this stress appears to be diminishing. If magma pressure overcomes the rock overburden in these more peripheral areas then it may happen here instead. Then it might jump back… As long as magma keeps filling the sill up, it’ll keep squirting out wherever’s easiest. However, to reach and supply these areas further away, more magma pressure would need to accumulate before the point of failure. If the infill of the sill is constant, then more time has to elapse before the next distant intrusion.

          If you’re wasting work time zooming about in Google Earth, the peninsula’s historic lava fields look far bigger compared to the recent flows, dated to a rough year or period (with a margin of error) and appear homogenous. However, multiple small intrusions and eruptions over decades could build up a lava flow field that, after centuries of erosion and plant colonisation, would appear no different from these historic ones. That’s not to say some impressive individual lava volumes haven’t been achieved, but I feel it’s the exception rather than the rule. The shields are impressive, but less numerous than the fissure rows.

          Are we witnessing the ‘classic’ eruption style on the peninsula where multiple individual dyke intrusions create a dyke complex along a fault system or graben over decades (not the best scenario for the people of Grindavik)? Or is this something new or rare in character? Is this unique to the Svartsengi system? This could be the case if it’s the only system on the peninsula to have this ‘elastic’ sill that breathes as the pressure changes. What if the sill fails in a different direction and forms a new dyke complex? What if it fails above the sill (constant expansion and contraction of a brittle rock layer induces fractures in time)? What if the sill inflation rate increases? What if the whole thing stops and we’ve got no regular lava fix? What will happen to Jesper? So many questions, and not enough data…

          Edit: I forgot to mention the recent current series of quakes on the southern area of the sill – around Thorbjorn/Grindavik. Is this a sign of the sill expanding horizontally, a sign of the sill pressure reaching ‘squirting point’, or a coincidental pattern from adjacent faults adjusting? Most likely the latter ones, but time will tell. A repeating pattern is something to keep an eye on – if that’s the case here. I haven’t been monitoring the seismic patterns before each eruption to see if this is the case, nor had the time to plot the data myself. If any one has, it would make for some nice discussion or a post of it’s own.

          • There does appear to be pulses of activity, but I also haven’t had time to look closely, so it’s pretty much anecdotal in scientific terms. It would make for interesting analysis though, if the data was as available as it was during Holuhraun. It would now rely on someone collating the data in real-time.

          • After Fagra’s attention cycle we’ve nearly forgotten that the first significant intrusion in Grindavik’s area was in spring 2020!

            This map shows the deformation in winter and spring 2020:

          • Sadly, I don’t have time to do any plotting at the moment, but I recently found out that Skjálfta-Lísa has an API that can be used to query data. This is a great improvement compared to parsing the week by week text format quake lists like I did before.


            You just submit a JSON formatted query using the POST method and get a JSON formatted response with the results.

          • Thanks for the API link Tomas. That’ll be really useful for extracting and constraining the results for a specific purpose.

            I’ll also use that along with an IFTTT trigger and use it for the Hekla (or any) alert service (I had it working a while back with a different data source but it expired). Not to mention the very out of date 3D map thing I did many years ago. Something to dig out and dust off soon.

          • You can look at the examples given in the API. It’s also possible to change the example requests and try out different settings. The json format is easy to read, structured plain text. Most programming and scripting languages have built in support or easily available libraries to read and write json formatted data. This is how you would format a request for all quake data for Reykjanes during 2023:

            “area”: [
            “start_time”: “2023-01-01 00:00:00”,
            “end_time”: “2023-12-31 23:59:59”,
            “depth_max”: 25,
            “depth_min”: 0,
            “event_type”: [
            “fields”: [
            “magnitude_preference”: [
            “originating_system”: [
            “SIL picks”
            “size_max”: 6,
            “size_min”: 0

          • @Thomas, Thank you. You’ve got six sets of coordinates for area. Does that mean that you are not limited to a rectangle?

        • The intruded magma can still play a role in one of the next eruptions. Sometimes new hot fresh magma first pushes out older magma.

    • Chad:
      I am not sure of your assumption that the underground region at Sundhnjukur is now almost completely loaded up. Can you please explain more on this rationale?

  11. Is Fernandina eruption already over or does it lack observation? Many volcanoes in the world, that are worth to be seen, have the disadvantage of little media coverage and likely also little ressources.

  12. A lot of vehicles leaving the Langihryggr camera stage right over the last 10 mins. Could they be evacuating the Blue Lagoon / power station?

    Has anyone seen any signs anywhere that something might happen imminently?

      • The traffic was the other side on Mt Thobjorn going north from the power plant / blue lagoon. I think directly behind this camera.

        There were a lot of cars for 10pm on a road that only goes there, now it’s been cut off by the lava. They were mainly leaving, although there were some going as well. Maybe it was a change of shift.

  13. I ran a bunch of inSAR pairs thru Cali’s Jet Propulsion Lab inSAR tool.
    The Spacing is 24 days, so spans two passes of Sentinel 1. The last pair is spaced 12 days, so one pass of Sentinel 1.
    I’m currently batch processing a series of 12 day spaced pairs.

    I’m the Guy that Carl called Icelands Highland rescue over a landslide in Askja Caldera a few years ago…. Because my Tracking beacon failed a few days in a row.

    • Stefan:

      I might be interested in your work (meaning the software techniques you are using to process the data) I did my own InSAR analysis using the Copernicus data back last year in the fall season time frame, but then things went strangely quiet when I was corresponding with a certain group of people (who are experts in inSAR analysis) Please keep us informed.

      Thank you

  14. The area of Grindavik/Svartsengi/Thorbjörn/Reykjanes was a hot topic 2020 on VC. It got a bit in Fagra’s attention shadow afterwards, but the events 2020 were likely first signs for what has happened since November 2023:

    January 2020 Carl wrote an article:
    Second article in February 2020:
    Albert on Reykjanes:

    The question arises: What has happened in the long time between March 2020 until November 2023 in the Thorbjörnes area? Was it quiet or did anything continue to go on without public notice? I can’t believe that this system was really dead for 30 months.

      • Either those prior were so called failed intrusions, or it was just doing the groundwork for creating the sill, magma burrowing into the rock at the base of the crust. I don’t recall how deep the quakes in those swarms were but I do remember there being some interest in a potential eruption west of Thorbjorn which was where most of the quakes seemed to congregate. That was mostly pre and post Fagra 1.

      • Covid only rescheduled Svartsengi and Grimsvotn, no one was expecting the magma to find another way out next door so it wasnt included 🙂

        • Nasty bugger, that magma. Always goes for the weakest point – it is just not cricket

      • May 2022 was another strong earthquake swarm of Svartsengi in Eldvörp region to the west of Thorbjörn:
        They analyzed that the swarm was “likely caused by magma intrusion”
        So we have the 1. intrusion in winter/spring 2020, the 2. intrusion in May 2022 after a break of two years (during which the initial Fagra eruption happened) and 3. the renewed intrusive activity since October 2023.
        This means that Svartsengi from 2020 to 2023 had two dormant periods: 1. two years from spring 2020 to spring 2022, 2. 17 months from May 2022 to October 2023.

        • Seems from those ranges there actually might be a connection between Fagradalsfjall and Svartsengi more directly, the inflation at Þorbjorn happened outside of when Fagradalsfjall was obviously active. It will be interesting to see if the current activity is indeed just going to keep going and terminate with a big event while Fagradalsfjall has had its time, or if the latter is just slower and still has a long way to go and we get a more episodic series of eruption and intrusion from Svartsengi. The last eruption at Sundhnjukagigar was much larger than anything we have seen so far and clearly needed more time to build up, maybe this is how that happens.

          Interupted magma flow for a more extended time is also a much more likely way to get the eruption to relocate. <1 month intervals we have now is probably too short to cool much so the rift gets reused but after a year it could be a different story.

          • I imagine magma has ascended / is ascending under the Reykjanes Peninsula in discrete blobs of varying size. Each blob finds its easiest route.

          • It is likely a partially negative relation. It’s possible that the systems act a bit independently, but also “steal” a portion of each others magma.

            I have the impression that Fagradalsfjall is like Hawaii’s Pahala area where the magma rises from the mantle and continues towards shallow reservoirs of Kilauea and Mauna Loa. Unlike on Hawaii the Reykjanes Peninsula has built a special “Pahala volcano” (Fagradalsfjall) where occasionally over Holocene eruptions happen. Fagradalsfjall is the direct “mantle to surface volcano” that erupts in some cases. Svartsengi and Krysuvik contrary to this get their magma indirectly from the mantle by the mediation of Fagradalsfjall’s deeper system.

  15. A lot of steam at the moment on the Langihryggr camera coming from the south end of the last eruption left portion of the screen). The ground is clearly a lot hotter here than elsewhere. And it’s where the previous lava flows were the thinnest. More than in recent days.

    Worth keeping an eye on it.

  16. I think it will erupt across Grindavikbaer road just north of the town. Sort of central to the original dike (but on the western edge of the most recent one). I say this because it seems to be an aseismic spot and there’s swarms of earthquakes either side.

    • Does the longer break lead to a larger eruption than we’ve seen until now? In some cases long breaks make big eruptions. The break until now doesn’t exceed the average breaks of December to February a lot. The recent intrusion may erupt together with the next one and cause a different volume and style of eruption than the first three events.

  17. How rapidly does magma evolve?

    Fagradalsfjall’s magmas were relatively primitive indicating a mantle source but lava’s from the current magma intrusion between Grindavík and Stóra-Skógafell indicate a magma reservoir in the middle crust. Has the reservoir been there since 2020 or earlier?

      • As far as I’ve understood Fagradalsfjall had more Magnesium content in its magma. This reminds to the early stage of Puu Oo and parts of Mauna Ulu eruption when magma came relatively unevolved from deeper reservoirs. Mafic magma like Basalt usually contains Magnesium as well as Ferrum (Iron), but primitive magmas have more Magnesium than ordinary Basalt.

        • Early stage of Pu’u O’o was actually more evolved, the MgO content was as low as 5.7% at the start and also during the fissure eruptions between Pu’u O’o and Napau in 1997 and 2011, as well as in the 1977 eruption which was further east of where Pu’u O’o would form. This was from magma mixing with stored rift magma, and many of the 20th century ERZ eruptions were very similar.
          The MgO got as high as 9.5% during some of the high fountains, and nearly 10% around the time Kalapana was destroyed. It declined after about 2000 to around 7% after 2011 but stayed pretty constant. After 2011 some magma began to bypass Pu’u O’o too and accumulate near the Heiheiahulu shield, which is where part of the 1955 eruption started from.

          2018 eruption MgO was as low as 4% for the first basalts in early May 2018 and the fissure 17 andesite was even lower between 2-4%, but on the other end the lava erupted from fissure 8 was as high as 10% MgO and hovered around 8.5% on average which is higher than most Pu’u O’o samples, and is much higher than any Pu’u O’o lava within 15 years before it.

          Fagradalsfjall had lava that was similar to the higher end of Pu’u O’o ranges. Svartsengi is maybe similar to the stuff that was erupted early on or during the other dike intrusions involving stored magma. Although because the relative numbers are higher in both cases the values are 1-2% high for the Iceland samples. But the ratio is about the same between early Pu’u O’o and the average as it is between Fagradalsfjall and Svartsengi, consistent with the latter storing its magma. I would like to see a number for the 2022 or 2023 Fagradalsfjall eruptions as those looked a little more evolved than the 2021 stuff so could be stored magma too or at least mixed.

          Pu’u O’o MgO variation over the eruption:

          Kilauea 2018 MgO variation:

          The lava erupted in September 2023 had 6.8% MgO which indicates the same stuff feeding Pu’u O’o in its final decade is probably still going. That being said the inflation rate of Kilauea has been very high in recent years probably as a result of decompression melting which has often lead to high MgO numbers in erupted lavas after a while (nearly ultramafic eruption in 1959 followed immediately by a big drain in 1960 then followed by more highly magnesian eruptions in 1965, 1968, Mauna Ulu, 1974), so this number might instead just reflect that 2018 didnt drain all of the magma under Halemaumau. The constant about 7% MgO of Pu’u O’o after 2011 might also reflect this too, the summit lake had become much more open by then and was the primary degassing source, so perhaps also indicated MgO was settling out in olivine, much of which probably got dredged up and erupted at fissure 8 to give it the high numbers.

          Not sure if Hector has already done this but plotting the MgO content of Kilauea from 1952 to present could provide a great insight to its activity and maybe some future extrapolation. There are many other variables of course but basalt is mafic and half of the word ‘mafic’ comes from ‘magnesium’, as such variations in this are probably a lot more influential than a variable of a few ppm of a trace element that is 10000x smaller in scale. Trace elements are like adding dye to water to see where it goes while the MgO is more like if 10% of that water was alcohol or something like that, makes a difference in the actual behavior.

    • Storage in the crust yes but the magma at Svartsengi is the same as at Fagradalsfjall in 2021 just a little more evolved. So it isnt magma from the ladt cycle as would be the case if it was a true central volcano. So I guess it is a polygenetic fissure volcano but not a permanent magma chamber. Which is good as it probably means to get a big eruption like the last eruption at Sundhnjukagigar or like Arnarseturshraun it will take longer. The bad news is that ‘longer’ will probably see lots of smaller eruptions,though maybe not so regular as it was so far. Maybe that is what happened last time, the start of the process to get a much bigger eruption. At 5 m3/s that is about 0.16 km3/year, so it might still take a few years to get a big eruption during which possibly nothing will happen. 5 m3/s is less than the supply seen in the eruption in 2021 though, so that would imply Fagradalsfjall isstill getting supplied just maybe not fast enough to sustain an open conduit like it did in 2021 hence the smaller fissure eruptions not lava shields.

      By evolved too, it is by an extremely small amount. 2021 lava was some 9-10% MgO which is very high, Svartsengi is 6-7% which for context is about the same as Bardarbunga (7-8%) and much higher than Grimsvotn (4-5%) or Askja (4-5%), the former of which in this very article being described as pure plume, perhaps needs to be fact checked… Its also very comparable to Kilauea and Mauna Loa which are variable between 6 and 12% each.

  18. Aurora nicely visible on the mbl Þorbjörn cam right now.

  19. I wonder if the current series of fissure eruptions has finished? While magma continues to move at depth, perhaps if has degassed considerably and no longer has the pressure needed to erupt?

      • I haven’f fhe toggiest idea.. I normally blame predicfive fyping for any misfakes.

    • Im not really sure if degassing to totality is enough to stop an eruption if there is continuous supply, its like a bucket being filled with soda water, it will still fill up even if all the gas escapes if you keep pouring more in. This might matter more in silicic sysyems where getting the magma to flow into a small crack is difficult but not at Svartsengi, where the magma is probably about as fluid as olive oil.

      At the depth of the sills under Svartsengi only CO2 will be escaping anyway and maybe not even that, apparently it starts to exsolve in a significant way only at less than 4 km. SO2 and H2O will stay in solution until near the surface so basically wont do anything until an eruption is near certain, maybe even being what drives the dike up in the final couple hundred meters.
      The magma erupted so far from both Fagradalsfjall and Svartsengi is similar, in 2021 the lava was 5000, 1500 and 1000 ppm for CO2, H2O and SO2 respectively. I couldnt find data on the eruptions after 2021 but everything I have seen otherwise about the magma is that the same stuff erupted every time so the numbers above are probably close enough. The intense lava fountaining when the fissures open at least indicates to me that the lava is gas rich.

      • It takes more than degassing to a halt to stop an eruption. Each time a blob of magma moves towards the surface, it goes down a pressure gradient. This causes decompression melting. Magma that was sitting in a semi steady state a few ks under the surface of the crust suddenly wants to erupt when it gets transported upwards. Even if it’s practically degassed to a state of equilibrium already.

    • Maybe the next fissure eruption will be able to concentrate on an erupting cone for some days. That’s the normal behaviour of eruptions. They begin as fissure eruptions that slowly reduce in length, first to several lava fountains and finally to a main vent.

      How long did Holohraun need to find a central vent? If I read the chronology correctly, the eruption began in August, but the first main crater/cone “Baugur” was built in October 2014. It was 100m high, so could be labled a volcanic mountain instead of a hill.

      • I dont know if Holuhraun really ever truely centralized, the fissure stayed nearly 800 meters long until close to the end, and it was only a few times longer than that at the maximum length. Not sure exactly why this was as Kilauea a few years later showed even higher average effusion rate and from a centralized vent. There were usually a couple of other vents along the 2018 fissure active alongside fissure 8 but very minor in comparison. Not like Holuhraun at all.

        Holuhraun also made a much larger vent structure. Ahu’aila’au is 30 meters tall and about 100 meters wide while Baugur is nearly 100 meters tall and probably 300 meters wide, and 1 km long at the base I would guess. Holuhraun had some very tall fountains in the early days going over 300 meters, while Kilauea maxed at about 100. I guess the relative scale is the same.

        Does make you wonder, if these two were not observed then just looking at the vent area no one would pick Kilauea as having the more intense eruption.

      • Ok so from actually measuring, Baugur is 250 meters wide, 630 meters long, and the crater is about 500 meters long and 100 meters wide, with long being along the axis of the fissure of course. Google Earth doesnt have great models and hasnt even got a terrain model at all for any of Holuhraun (or really, almost nothing new volcanic that is formed in modern time…), so I will go with the ~100 meters still although I would expect it to be less than whatever it was a decade ago due to settling and cooling.

        Ahu’aila’au is 220 meters wide and about the same length, it does have the deep channel and then a wide shield on the east side though which makes it hard to determine the base of the cone on that side. The crater is a little under 100 meters wide and about 130 meters long. Apparently the peak height was 55 meters although like Baugur it is probably settled and lower than this now, my guess is probably some 35 meters.

        So actually Baugur is a bit smaller than I thought and Ahu’aila’au is a bit bigger. The distinct crater at the south end of Baugur is very similar in size to the crater of Ahu’aila’au, so its pretty easy to get a sense of them side by side.

        For comparison the two largest craters of Laki, where the eruption started, are both about 370 meters wide, and the tephra cones about 900 meters wide. These two erupting would have been a sight to behold, with constant fountaining as high as the very tallest that we saw in 2021 but probably going for weeks on end.
        Most of the Laki craters further towards Vatnajokull north of Laki mountain are about the same size as Baugur or Ahu’aila’au though, ~100 meters wide, so the majority of the eruption duration might have looked pretty similar just with more vents open. The stage 1 vents really did a lot of the eruption early on. Still having the two largest effusive eruptions in modern time as a proxy for the average eruption rate of Laki is pretty extraordinary for how long it lasted.

    • Where has the gas gone in that case?

      Is it still trapped in the headspace of the dike?

      It can’t have got into the atmosphere, otherwise a massive gas signal would have been detected when the system *wasn’t* erupting.

      • I guess it is slowly diffused through the crust, but that would only apply to CO2 and probably not in a meaningful way without an open vent. But I think this might be one of those things which sounds like it matters but doesnt really at least not on the timescales in question.

        Given that Kilauea has a comparable if not even higher magma supply than Svartsengi and a shallower storage system, yet seems to be absolutely closed to any meaningful degassing unless it is erupting, probably says enough. The only volcanoes that seem to be able to degas in high volumes without erupting are stratovolcanoes, maybe because of the composition of loose gravelly rock they are mostly made of. The only example of a mafic caldera or shield degassing in a significant way without an associated eruption is Grimsvotn a few years ago. Taal is also a heavy SO2 emitter but its not clear if that is shallow magma or because of intense hydrothermal circulation, which js a different thing.

        • Should say the only example I can think of, there are probably more 🙂

        • The smoke is from the peat/moss burning at the edge of the lava field and the vertical plume that occurs around 16:00 coincides with a change in the wind direction, possibly facilitating the increase in plume height in a slack wind field. I can’t see how CO2 exsolution from magma at 3-4 km depth can account for this as, Mike mentioned above, the CO2 plume would have been detected by measuring equipment. Plus, there would be a delay from emplacement time (~16:00) and the gas escaping from the surface.

          • I have studied the timelapse quite a bit. You can even go full screen and enhance it 50% and move main area of interrest. From watching it over and over it seems quite clear that this is not peat/moss burning. If you have seen some of the many drone fly-overs done since you will recognize many of the smoking areas are well within the lava field. See esp. from 1.30 till 2 mins out and this coincides with the intrusion. I think Occam’s razor would agree based on information at hand.

          • Apologies, I had my bearings wrong thinking this was looking East from Thorbjorn (and didn’t read things properly on the webcam) – It’s actually looking West from the ridge to the South of Fagradalsfjall. This puts Sýlingafell at the centre of the camera view. The smoke plume you mentioned isn’t smoke, it’s steam from the power station that’s situated on the other side of Sýlingafell from this angle (hence why it’s illuminated at the start of the video in the dark). There are high plumes of steam earlier in the day (~11:00 am) that don’t coincide with the intrusion. I don’t see any evidence of surface-related activity from the intrusion at ~16:00 pm.

          • Hello again.

            (Replying to your latest answer) If you look at the same timelapse for Svartsengi it is easy to distinct the difference btw steam and C/S based smoke. Steam dissiapates quickly, while the smoke I’m reffering to follows the ground more and creates a long band of lasting smoke. Steam doesn’t. Not to say there isn’t steam involved in the lava field too.


            Timewise I find it fits very well. But again, one tends to try to confim what one thinks. It is only when the dike starts to near it’s breaching point that we see large eq-swarms. It is not unlightly that increased smoke is happening right before this. We are only talking about 40-50 mins. pre the eq-swarm.

            I will check this on timelapses post the next intrusion. It *could* be a precursor.

            I really liked your graphics on the mechanisms at play below in the comments. And I hope you can make time to cover it in an article on VC.

            I have some questions and comments to it;

            1. If one considers the tectonic rifting part of this activity to be wedge style from deep -> surface it would not be unthinkable that the rifting is more even – – or different, and precedes surface rifting – at depth than at surface where we have observed that it happens at the time of eruptions as if the crust snaps. We can’t measure this the same way at say 5-10 km depth, but if it precedes (it has to fill voids from vertical rifting at all depths) failed eruptions also can be explained. If there is a fairly sudden rift episode at depth this would relieve pressure and create more underground space for the magma, thus causing the magma to retreat. What are your thoughts on this?

            2. However, why doesn’t it seem like the dike drained after this? There is some evidence that the ground deflated slightly after the episode on March 2. (Eldvörp ELDC, Grindavik GRIV and Thorbjörn THOB GPS for instance) supporting a) magma going into the dike and/or b) tectonic expansion of the sill. However (with the grapics in mind), sholdn’t the dike also be drained back somewhat if the grapics depicted the situation? Pressure clearly subsided as the eq’s decreased. The sill(area) deflated somewhat. So why is there no/little indication of magma going back into the sill? It kept smoking. It is only the last few days that smoke (and steam) has decreased arond the greenhouse N of Grindavik. The same at the foot of Stora-Skogfell. From the last eruption it is observed that the magma is very liquid. So it should easily drain back to the sill based on relifed pressure, viscosity, gravitation and tectonic rifting. This has been observed in failed eruptions in craters in areas where tectonic rifting is known.

            So what am I missing here? Since I am not a scientist I simply ask if there is a known mechanism in place or if this is (fairly) unknown territory. Out of never-ending curiosity and the never-ending need for explanations.

            If I were to experiment (to observe) with the scenario (let’s say with water) I would propose an upside down kitchen water lock to explain it (analogy). Because then the fluid would fill the higher up dike from the deeper sill wo returning once depressureized. So maybe under Hagafell? Or maybe totally wrong bc of different dynamics at play.

            I hope you see this – and can make some sense of it.

            All the best from Norway.

          • ADAV, thank you for your measured response. Before dismissing your observation as I did, I should have requested a screenshot of the area you were talking about. There are so many different aerial plumes of different types in the shot and I’ve obviously picked up on the wrong one. It’s been one of those weeks, and the endless winter of viral respiratory infections affecting the kids and I and resulting lack of sleep hasn’t helped. If you want to screenshot and highlight the area in question, I’m happy to continue the conversation on the new post that’s up.

            If you can email the VC mail with your questions, I’ll add them to the list to look at and try and answer as part of it, if I don’t get time over the weekend to answer them.

    • Won’t be meaningfully be degassing down there, especially on the timetable of weeks. If it shuts down, that would be because supply slows down.

  20. Hofsjökull with a 2.1 quake ad 0.1km depth = a typical hydrothermal quake? Katla does hydrothermal quakes occasionally at the same 0.1km depth.

    A different topic: Which of the Eolian volcanoes apart from Stromboli and Vulcano are able to erupt again?
    – Lipari was clearly the last active island during Middle Age 1230
    – Alicudi, Filicudi, Panarea and Salina no eruptions during Holocene
    – offshore Marsili Seamount last eruption 3000 years ago
    This looks as if only Lipari is a vital volcano. The other volcanoes may still surprise in future, but unlikely.

  21. I think Fagradalsfjall is different in that it formed directly through the crust in a more vertical fashion from the upper mantle, which is why it was able to sustain its eruptions as they were constantly supplied. I do not recall there being evidence for a nearby sill supplying the dike here.

    Svartsengi is more indirect, it has first formed an intermediary sill which then has to inflate to overcome the pressure threshold to push out eastward into what is obviously a more efficient eruptive center (the dyke) which is probably in more brittle/ductile crust than where the sill itself is. It’s just entropy, second law of thermodynamics.

    It could well be the exact same base magma from the upper mantle so the chemical differences would be mostly caused by crustal contamination and storage time/fractionation.

      • Thanks. Interesting that historic Svartsengi (and also Reykjanes) lavas have the lowest MgO and K2O enrichment of the peninsula whilst Fagradalsfjall had by far the highest.

        It suggests magma spends more time in the crust westward towards the MAR than inland. Why this may be the case in anyone’s guess, perhaps there is less tension in the crust or less lateral movement of the plates the further west you go, so at Fagradalsfjall (or Brenninsteinfjoll) the magma comes out all at once in one long burst (the tectonic swarm was at it’s most violent in the east of the peninsula), whereas at Svartsengi it can gradually store up before being released when the pressure is reached.

        • Andy:

          I was reading a paper about the subsidence of Svartsengi (before the 2020 and onward inflation episodes) which is basically contraction of the crust due to cooling, but when I got to the borehole profiles (which are accurate) I realized that we are looking at a very complex situation. There have been many episodes of lave fissure and shield eruptions, and this emplacement of lava makes it very complex to unravel how things will proceed.

          I think most of us are looking for simple explanations or succint conclusions, but the borehole data shows us that making simple models might be doomed to fail. Right now, all we can is watch and see what happens in March and April.

          • We only have the historical record of one Fires cycle and relative good data on the previous, ancient one. This is a small sample to create a model about the timeline of a Fires cycle. It’s both possible that all Fires cycles are somewhat alike and that all Fires cycles are different. Are there certain types/classes of Fires cycles that occur repeatedly again? Is each Fires cycle unique in its chronological sequence?

          • Very true, it is a small sample size. If the crust was contracting beforehand then this has been a very sudden process!

    • Lava accumulation has reached the level of February 7th. Episode IV (would get the title “A New Hope” in Star Wars) can begin anytime:

      • Here’s the English version. It’s currently broken on IMO’s English site but I tweaked the URL until I guessed the right source domain.

  22. For Chad and anyone interested. This is the MgO weight percent content of Kilauea lavas per year 1952-2018 (data from GEOROC):

    • Is this the whole rock percentage or just the glass/melt composition? I wasnt aware it was close to 20% in 1959 and in Mauna Ulu, if that is the glass composition then the melt might actually be genuinely ultramafic or very nearly so. I would assume it to be from olivine crystal load though.

      The general declining trend looks like it is showing the growth of a large body of liquid magma, where the olivine can settle out and bring the MgO content down. Of course this was obvious with what happened in 2018 but post 2018 lava is still the same percentage as the later Pu’u O’o stuff so evidently the chamber is still largely intact despite what it appears. I recall you once showed a chart of DI events too and they were rare before the 2000s abd became very common in the interval after 2011 when the MgO stabilized at about 7%. Your idea of DI events being from the caving in of walls between sills or similar processes fits very well with this.

      As a side, from the known deflation scale of the eruptions and the February intrusion each microradian on the tilt seems to very approximately correspond to about 1 million m3 of magma moving in or out. So each DI is basically a few million m3 of material draining from the chamber followed by rapid resupply of about the same amount, or a little more on average right now. This is only at UWEV, the others will have different ratios as this is only a correlation not a direct measurement, but it does help understant how dynamic it is down there even if there isnt any surface lava 🙂

      • They are all whole rock samples. There are few glass samples from Kilauea lava flows. Although some lavas are largely aphyric, like many Pu’u’o’o lavas, I think. Kilauea Iki had picritic lavas though, with a lot of olivine. There are glass samples from Kilauea tephras, though. The biggest eruptions like 1790 and Kulanaokuaiki-3 have evolved glasses, like 6 wt% MgO for the K-3 eruption. However some of the Keanakakoi Tephra eruptions are quite primitive.

        A so-called Keanakakoi unit 20 eruption (I think from the collapse around 1660 AD, possibly related to the large eruption in Kapoho) averages 9.7 wt% MgO glass composition, this is a lot. And apparently, the most magnesian glass of Kilauea is from the Puna Ridge at 15 wt% MgO. For reference, 10wt% MgO tends to be the upper-end primitiveness of mid-ocean ridge whole-rock basalts, the lower end being around 5 wt%, although those are quite rare. Fernandina usually erupts 6 wt% MgO, Sierra Negra 5 wt% MgO, and Katla 4.5 wt% MgO whole-rock basalt. Most people are not aware that silica is only a good indicator of how evolved magma is for the late stages of magma evolution. Early on during magma evolution, magnesium gets rapidly depleted with only a slight increase in silica. There is often a middle stage is which magnesium, calcium, and aluminum are depleted, whilst titanium, iron, potassium, sodium, and phosphorus increase in concentration, but silica often does not shift at all. So basalt can evolve a lot and remain basalt. Past a certain point, it starts to build silica really fast. High-magnesium andesites that are almost ubiquitous in volcanic arcs to a variable concentration in the melt do not follow these rules, though, and are inherently silica-rich magmas.

        • If the case that 1790 was a major caldera collapse is correct then that indicates the magma chamber that collapsed must have been long lived by that point to allow so much olivine to settle out. Being at the summit though the temperature and physical characteristics likely wouldnt change much though unlike at rift magmas with similar degrees of evolution, still 1200 C or more down there, im not certain the normal MgO-temperature correlation of lava would work for that situation.

          Its a shame the 18th century record of eruptions along the middle ERZ is now permanently erased and the only stuff we have to go on are old photos from 1954. Most of the flows of that era seem to be huge lava flood type eruptions, not particularly voluminous but extremely intense. There was also Heiheiahulu which probably ended in about 1760, by analogy to Pu’u O’o this might have been when the 1790 magma chamber begun to coaggulate. It seems to have ended the same way as Pu’u O’o too though less dramatically. At least visually some of the flood lavas where Pu’u O’o is now look younger than the Heiheiahulu flows.

          I guess that is the price of having such an active volcano, Pele hides her secrets fast and very well…

    • Thank-you Hector! The MgO content was during Pu’u O’o above 8% 1985-1998, later it decreased below 8%. Does this mean that magma became evolved to classical Basalt after 1998?
      The end of 1950s and end of 1960s had peaks of very high MgO between 15 and 20%. The 1950s peak was related to Kilauea Iki, while the second peak was around 1968 with eruption at the Hi’aka craters and later shift towards onset of Mauna Ulu eruption.

      • My theory is that after the 90s was when Kilauea began to form the magma body that created the 2018 eruption. Before 1924 there was a similar situation that HVO referred to as ‘source 3’ for deformation. This is what created Mauna Iki presumably, and was basically just an enlarged conduit of Halemaumau probably at around sea level. Its not entirely analogous as Kilauea back then was not as active as now in effusion rate but the 2011-2018 era and 1918-1924 era both had in common the existence of a large actively circulating lava lake. The exact same thing probably happens at Nyiragongo and leads to its flank eruptions although that is just theoretical and the two volcanoes are structurally very different of coyrse.

        Basically I think when the chamber becomes homogenized and is an actual large volume of liquid and not just many layered sills, it allows olivine to settle out which reduces the MgO content. The fact the Galapagos volcanoes have relatively low MgO and have got large permanent magma chambers of significant dimensions would be in line with this.

        The next decade will be possibly very telling at Kilauea in regards to what it might do in the future beyond. Since 2018 there has been very strong inflation which would hint at increased supply but the magma of all the post 2018 eruptions has still been the same as it was before. If the magma becomes more magnesian over the next 10 years or so then maybe we are just at the start of a new chapter like in the 60s. But if all the fresh magma erupted in the coming decade is still the same lightly evolved magma erupted since 2011 then maybe 2018 was not the final show. Im somewhat inclined to go with the second option personally.

        • The development of MgO content 1950 to 2018 looks like a cycle with new primitive magma during the beginning and a tendency towards evolved ordinary basalt. Kilauea Iki and the 1968 eruptions were the double onset of a 60 year “Fires” period. There was an ongoing evolving process of magma. 2008-2018 was the final phase with the most evolved classical basalt.

          This cycle reminds to the onset of the current RVZ Fires with MgO rich primitive magma at Fagradalsfjall as RVZ’s Kilauea Iki and now Svartsengi with more evolved basalt.
          It’s possible that previous RVZ cycles made intrusions at Fagradalsfjall without eruption.

    • Thanks for the link Randall! Why is the Kalium content higher than 800-1240?

      • Apparently, the lavas are more alkaline than in the last cycle. The first Fagradalsfjall lavas were very close to the previous Reykjanes Fires eruptions. But early in the eruption, they got gradually more alkaline (richer in incompatible elements like potassium). I remember an article that looked into this. It seems the alkalinity has stuck and continued into the Sundhnukur Fires.

        • What is a possible reason for more alkaline basalt now? Does this K content indicate that the current RVZ cycle belongs to a different type of Fires than the previous one?

          • Could be magma from central Iceland. Central Iceland is more alkaline.

          • Don’t know but the current eruptions have occurred fairly shortly after twitches at Vatnajökull and Myrdalsjökull.

          • Does this imply that Myrdalsjökull can be involved in future activity? Eyjafjallajökull 2010 was relative short before the onset of RVZ’s activity 2020. During the 10th century Katla erupted both ordinary and extraordinary (Laki) after RVZ had done the first eruption at Krysuvik.

          • Did you mean Katla will erupt a close succession to activity on the RVZ or that Katla can feed the RVZ? Volcanic activity can hop from the system to a nearby system through time as the stress field changes across Iceland to accommodate the uneven plate spreading. It’s not definite though as there are many variables at play. Magma from Katla wouldn’t be able to influence the RVZ directly, it’s too far away. Katla was responsible for Eldgja in the 10th century, not Laki (Grimsvotn).

      • Volcanophil:

        I honestly do NOT know the answer to this question, but I see that Héctor Sacristán has chimed in. I appreciate his response.

    • I teresting so the MgO is not quite so high as I thought, a little under 8% for Fagradalsfjall (that I presume is all of the eruptions if it was done this year) and 6% for Svartsengi more or less. The stuff from the southern fissure in January being both more crystaline and evolved than the northern fissure, as well as its opening way later than the north fissure, that supports it being November dike magma pushed out by the newer intrusion.

      Still even the Svartsengi lava is significantly more magnesian than most of the Vatnajokull volcanoes, the south fissure with the most evolved magma still has more MgO than both the Grimsvotn 2011 and Laki samples, (5.8%) and more still compared to Askja and the volcanoes on the south coast (<5%). Bardarbunga is highly magnesian though, being around 8% in Holuhraun, which is expected of it being the plume center.

      • I was wondering if 3 – 4 years was long enough for the MgO levels to drop from 8% to 6%. If yes, the magma erupted at the end of 2023 and earlier this year could have ascended at the start of and during the volcano-tectonic episode which started at the end of 2019. If not, it has percolated up earlier.

        • Considering how variable it was for Kilauea I think it is reasonable for 2% to drop in a few years, if the lava is very fluid and is allowed to settle then the crystals will fall out, its like if you sprinkle sand into honey. Maybe if Svartsengi did erupt in 2020 then it would be higher MgO but the times the supply stopped let things settle. Maybe if magma keeps flowing in fast then the MgO could go up but probably not.

  23. Looks like the Big Island is showing some signs of the building pressure.

    2024-03-13 11:39:19
    2024-03-13 11:18:00
    2024-03-13 11:16:00
    2024-03-13 10:28:02
    2024-03-13 09:55:43

    • Kilauea starting to swarm on the SWRZ again. The GPS is showing that about 60% of the early February intrusion has been recovered, and both SDH and UWEV tiltmeters are recording inflationary tilt that is now much more visible without the intrusion offset.

      I had earlier predicted another intrusion in May, and probably more likely to erupt this time around with the last intrusion taking up a lot of space underground. But it looks like things might be moving faster by

    • Maybe April is a good time for a volcanic Easter Egg of Hawaii?

  24. I hope a new magma pathway is not being created past the south of Thorbjorn to the north of Grindavik, but we keep seeing 4/5km quakes cluster here.
    We will soon find out if the old pathway has sealed itself in that last magma run, or if that material will be forced out.
    Geologyhub speculated the recent small intrusion was gas rich hence the severity of the earthquakes, but not enough volume to erupt. This meant that pressure would have dropped in the sill due to the gas release. I suppose it’s sound reasoning but it looked like the GPS barely flinched and it now seems to be levelling off.

    • There have been quakes just to the east of Grindavikurvegur road for some time now. I watched them in that area before the last eruption and then before the newer intrusion a nd multiple quakes occured along that road 43. each time though the action was well nE of there. Something though is causing those quakes and too many in that same area to ignore in my opinion. Time will tell but I wouldnt bet against that area being a focus of another eruption one day.

      • The time to worry is when they get shallower, with stacking. But even then, they could be triggered quakes. We are probably seeing the crust responding to uplift at Svartsengi, etc.

        To date, eruptions have been preceded by earthquake swarms (albeit with short notice).

        • Also, if the eruption is not between Stóra-Skógafell and Hagafell, magma probably still has to work harder to break through so there is likely to be an earthquake(s) in the swarm great than or equal to 3 or 4.

          But all speculation until there is an eruption.

        • See my comment further up the comment tree where I discussed/speculated on these quakes:

          I would put them down as crustal quakes with normal faulting (don’t mention the beachballs); the brittle rock responding to the sill pressure reaching a certain limit below them. What the sills limit is, and how it changes from each inflation event, is something we can only hope doesn’t involve seeing Grindavik wiped out. As long as it keeps erupting roughly where it’s been mainly focused, Grindavik may have a chance. I’ll post this graphic again from the recent paper on the Nov 23 main dyke formation, in case anyone’s missed it and it helps visualise the possible system below:

          Although useful for this purpose, I think the graphic could be expanded on. Especially with the representation of the magma domain/sill and including a better depth scale showing how deeper sourced magma arriving around 2020 can be linked to activity in this and Fagra upper-crustal systems. I’m hoping to make some quake plots soon with the data from the IMO quake API that Tomas kindly linked to further up the tree. It’ll probably form the basis of a post on the Reykjanes 2020 – onwards episode (and yes, it’ll cover a time before that date). I’ve still got a few published papers to read through before that can be written though as well as make some graphics up to help visualise the things I’m discussing (that may take a while with my current workload). It might also be a good time to email me at the VC email address ( with a few key questions or concepts that you might like to see more information on or explain in more detail or any useful papers/graphics/plots that you find in case I miss a key piece of information.

    • I’d rather assume that the not erupting intrusion of early March will be part of a more complex eruption in future. Maybe an eruption with two central vents.

  25. A new eruption has started at a volcano called Tinakula (in the Solomon islands). Looks like Stromboli’s long lost brother.

  26. Is IMO taking a well-earned break or has there really been no earthquakes since 15:21 today on the Reykjanes Peninsula?

    Small drop in inflation at Svartsengi on the 8hr and 4hr plots. Too soon to see if significant.

    • Merlot:

      I would not be that concerned about the IMO office being derelict. They actually do a really nice job, all things considered. I have carefully watched the IMO webpages, and they do finally update the earthquakes after a time delay.

      I know that it might be frustrating, but we need to cut them some slack.

      – Randall

  27. Something has changed on the Grindavik DAS. The plot has gone all blue on the bottom half and redder on the top half. What could cause this? Has the cable been severed?

    • Don’t know. It looks upside down. Or someone has change the vertical scale?

      • Joking aside, it looks like it is quiet round Grindavík at the moment.

        Combined with the lack of earthquake activity, not sure that is a good thing.

        • A scale change might make sense. Although the noise of the sea usually shows at the bottom of the chart, and I’m not sure the sea has gone that quiet. There are also some signals still on the bottom half, which suggests the cable isn’t completely broken. But it does look to me that the cable may have been damaged about half way along. Could this be the crust tearing and damaging the cable? Could it be heat? Could it be a digger? I suppose all are possibilities as well as many other mundane things.

          • Earthquakes today around 7 o’clock Icelandic time indicate magma flow from Fagradalsfjall’s deep sources to Svartsengi.

            Fagradalsfjall’s earthquakes:
            06:28 at 7.7 km depth
            07:20 at 7.6 km depth
            07:30 at 7.6 km depth

            6:31 at 7.3 km depth
            7:03 at 3.5 km depth
            7:11 at 2.6 km depth
            7:12 at 3.2 km depth
            7:27 at 1.5 km depth

          • The cable is OK. A signal travelled down it from c.08:15 onwards. Don’t know what caused the signal but probably man-made, being a test or a vehicle.

    • IMO has compared the intrusions to Krafla:
      ” In the Krafla unrest magma always intruded into the same dike, but at different scales of magnitude.”

      IMO’s comment to the graphs : “The image shows the interaction between magma intrutions and ground uplift in the center of the Krafla caldera. The graph below shows the elevation of measurement point within the Krafla caldera and the graph above shows the distance from the caldera to the dike formations.”

    • Either the Atlantic has gone as still as a lake on a calm day, or they have enabled some background noise cancellation algorithm to remove the noise generated by the ocean waves. The M1.9 at Reykjanestá at 09:14 looked really clean in the bottom noise free part, but was not visible in the noisy top part.

    • translated with Google to another Viking language …

      They say that more magma is needed than before to make a new eruption. Scientists of IMO assume that the system of dikes has changed. This increases uncertainty when and how the next lava show is going to happen. They still expect the next eruption to occur between Stóra-Skógfell and Hagafell.

  28. With all the talk about Kilauea this seems very appropriate, first seismic swarm following the recent intrusion. The SWRZ connector is active discontinuously along its whole length from the south caldera to the start of the Great Crack, showing it is apparently completely open. The February intrusion was the first of many it seems.

    • There are swarms over the past 2 weeks south-west of Eldvorp and south-east of Thorbjorn. I think it’s trying to find and alternative pathway to intrude but I hope I’m incorrect.
      The random quakes towards Fagradalsfjall and off the coast have been a strong sign that the crust is pressurized and that there won’t be long before the next magma run.

      The article does mention several earthquakes >M2 in the area and I don’t think i’ve seen any mind…

      • A professor thinks the next eruption will be at Eldvörp and in the autumn. Inflation is apparently moving westwards.

        (Think I must need new specs. Inflation is ongoing and seems largely unchanged to me.)

    • Is this due to the stress on the crust? 40km seems like it should be aseismic.

      • They are long-period earthquakes so probably related to gas/fluid volume changes in magma, perhaps magmatic gas being released through some cracks or some other similar process.

      • 40 km is the bottom of the lava pile, where the volcano sits on the oceanic crust. It is depressed underneath Mauna Loa because of the sheer weight of the mountain. Go towards the coast (any direction) and the layer becomes less deep. This connecting layer is a fairly frequent source of earthquakes as the crast adjusts, but it is also something that rising lava has to get through which seems to be happening now


    “They predict that the upheaval at Grindavík will end in late summer, that is to say in four to five months.”

    That’s bold, but they have a “track record”.

    Andy; you wrote, “I think it’s trying to find and alternative pathway to intrude but I hope I’m incorrect.” That’s what Shawn Willsey thinks.

  30. The starship had another test ride. It again ended in failure but a better failure than last time. Launch worked, booster landing failed badly and re-entry was a bit of a disaster. But at least it got off the ground safely this time.

    • Technically, if this was any other rocket it would be successful already, it could have but a payload into orbit if the flight plan was changed to a full orbit, the rocket surviving and landing to be reused is not really a requirement for function just for practicality and long term cost. Its actually remarkable how well this went after only 3 tests. We dont get this sort of observation from anywhere else except maybe RocketLab but the vehicles in question are obviously quite different.

      The re-entry footage was incredible too, getting to see it so clearly. Hopefully that is something we see a lot more of. Maybe Dragon will start getting re-entry footage now too 🙂

      • Sure. But the starship is not designed to just put stuff in orbit. The fact that the engine did not re-ignite to facilitate the re-entry is a bit of an issue. And the booster landing is a Musk special, and makes the project economically feasible. It is not an optional add-on, although it might be for other companies. So, a way to go and not human-rated yet but they seem to have fixed the rocket problem where quite a few did not fire on take-off last time.

        • I think Space X is following more of a HALT (Highly Accelerated Lifetime Testing) approach? It may have been that some systems were/are intentionally being pushed to their limit, which can only be determined by failure?
          But that’s only a guess. I wasn’t there.

          • That is what I have seen it described as too, rapid ineration after testing to failure. Its basically trial and error byt to be honest I think this way makes the most sense, its what got Apollo of the ground

          • Just to add, extensive testing like this is routinely done on virtually all components and equipment long before launch.,,but there is nothing like the “real thing”.
            Here on planet Earth, the vibration tables in particular are impressive to see. The shake tables that the lab I worked with were modulated by a bank of over 100 Macintosh audio amplifiers that produced enough power to drive giant high-speed actuators that could shake the table at variable frequencies and amplitudes. In such a test, you intentionally shake the test object until it breaks, which is different that trying out a new design and seeing if it survives under specified conditions.

          • For manned tests yes but not hardware development. And nowdays spacecraft can only be human rated in their final itteration. It took Falcon years of reliable launches to be human rated. Apparently SLS gets to skip that step and it did workfirst time but I would still rather go to space on a Falcon personally, its like being the first passenger on a plane you would hope it has flown more than once 🙂

        • I think there are some other aerospace startups with intended propulsive landing but none are commercial yet. ULA Vulcan is supposed to put the main stage engine in a heatshield for traditional re-entry and reuse but the main body is expendable. RocketLab is going for full reusability of the first stage which is good as they operate in a market SpaceX mostly avoids.

          I guess the parallels between SpaceX and Tesla are pretty apparent, both were the poineers doing something that was considered impossible, and in both cases it is the old guard (be it large car companies or aerospace) that seem to be in stubborn defiance of the change as a whole while now a swarm of startup companies are trying their hands at new ideas now, and the likely trend is an imminent rapid turnover of hands, a mass extinction if you will. Its an exiting time to be alive thats for sure, it sounds like what the Apollo days were always described as except this time it isnt just an ego contest and there are actual beneficial goals.

    • Quakes of this magnitude are very typical of the MAR in this region of the mid-Atlantic.
      I would guess at least 3-5 happen every year. This quake was from normal faulting, meaning extensional strain has been released. Of note is that MAR has a remarkable consistency in firing off M6 quakes all along it’s length in the Central Atlantic, which may indicate the crust over the fault has a uniform and finite limit before it snaps. Given that the crack that forms the MAR is made from previous magma extrusions, it makes sense that the basalt would have a similar tensile strength.
      That’s not to say that bigger quakes can’t happen…I think an M7 is the tops ever recorded along the MAR. But then again, that’s only during the age of seismometers…anything that happened before then (except Iceland where humans lived) would likely have gone unnoticed.

      • Interesting that the divergent plate boundary causes so major earthquakes! Usually I rather associate subduction zones (f.e. Indonesia) and strike-slip faults (California) with major earthquake risk.

        How often do eruptions happen along the MAR? Do they regularly observe volcanic tremor that would indicate volcanism in the deep ocean?

        • I don’t think MAR eruptions can be detected unless a seismometer is in the immediate surrounding, and that has obvious difficulties. Most likely we will continue to know little of world’s most voluminous form of volcanism.

  31. Its kind of hard to see on the public USGS maps but the Pahala swarm actually looks like it follows one of the fault lines that was active at the surface during the intrusion in early February. They are not exactly aligned but very similar otherwise. Perhaps the Pahala swarms we see are the result of magma intruding into a fault line that is structurally the same as the decollement fault which is related to the seaward slip of the island in that direction?

    The magma might be its own thing or it might be magma that would have gone to Mauna Loa but was diverted away by the topography of the base of the crust. The same article I linked to talk about the MgO content of Kilaueas lava during the Pu’u O’o era also mentioned that some of the lava erupted over the turn of the century had a subtly different composition closer to Mauna Loa lava though without absolute certainty of this identity to be fair. Its also pretty clear that the intense seismicity under the Pahala area is a Kilauea thing, being that it intensified enormously after 2018 and the Mauna Loa eruption in 2022 had no affect at all. But all these questions are still unanswered really. If there really are sills of the dimensions suspected down there then each one has the volume of magma to do a Laki, potentially anyway. Kilaueas caldera could maybe contain 1 km3 before overflowing, and the level it is at presently is high enough to start flank intrusions. Mauna Loas caldera is already filled to overflowing at each end and the thing is basically one gigantic fissure volcano now. So a big surge isnt going to have much resistance…

    • Next Mauna Loa eruption could be a summit eruption with long duration and relatively low rate.

Comments are closed.