Laki deconstructed IV. A Timeline for Destruction

Before I start I would like to point towards the previous 3 parts. It might be a good read for those who missed them when I first published them.

Prequel Laki deconstructed: I. prequel

Part 2 Laki deconstructed II. Anatomy of an Eruption

Part 3 Laki deconstructed III. Grimsvötn and Beyond

The pre Skaftár Fires eruption at Grimsvötn

Two weeks (15 May 1783) prior to the eruption at Skaftár a Brigg called to port at Höfn; the Captain dutifully wrote in his log that there was an ongoing eruption at Grimsvötn. There are few things we can say about the eruptions at Grimsvötn during the Skaftár Fires. What we can say about the eruption prior to the Skaftár Fires is that it most likely was a minimum of a VEI-2 and a maximum of a VEI-3. If we compare to contemporary eruptions it was most likely an ashy VEI-2. Anything smaller would not have been noticed, and might even not have been able to break through the ice above the caldera, and a larger one would most likely have been written more about.

The same thing goes for all of the activity at Grimsvötn prior, during and after the Skaftár Fires. The eruptions from Grimsvötn back then were fairly small compared to the activity that we have seen during the last 50 years.

The Skaftár Fires Timeline

Úlfarsdalur, the scenic starting point of the Skaftár Fires eruption. Photograph taken from

The eruptions started at the SW part of the Grimsvötn Fissure Swarm at Úlfarsdalur and extended all the way up to Sídujökull.

We know that there was a minimum of 10 eruptive fissures along this line; they are subdivided into the SW and the NE fissure rows. The dividing point is the old Lakí Mountain that is a hardened leftover from previous phases of volcanism. As such there was no erupted lava from the Lakí Mountain, even though it has a few tuff cones as sub-terranean material was blasted out.

The fissure rows contain five proven fissures each, even though there is a high probability that the fissuring continued further to the NE under the then present Glacier. But, as glaciers move they bulldoze away any proof of such a thing having happened. As we get closer to Sídujökull the evidence is more and more destroyed, and the timeline becomes a lot less certain.

I should here like to point out that there are temporal changes in the eruption as it progresses from SW to NE. From the first fissure we have tremendous output and a much higher rate of activity then what happened as the last fissures opened up. Also there is a temporal shift during the lifespan of the fissures. As they opened up they normally ejected about half of their total output within the first few days. There is also a third temporal shift at play during the fissure eruption, and that is that the fissures opened up with ever increasing intervals. One should also note that the fissures stayed open throughout the entire eruptive period, so fissure number one was open a lot longer than the last fissures.

Just to point out what we are talking about, as the first Skaftár fissure opened up it erupted about 2 cubic kilometers of lava within the first 48 hours of its existence. During the rest of the eruption it effused about 2 to 2.5 cubic kilometers of lava. That gives that the first single fissure erupted roughly 25 percent of all the lava during the Skaftár Fires, and that the same fissure erupted 12.5 percent within 48 hours of the start of the eruption. If we then use an inverted logarithmic scale to the fissures we find that the last fissure erupted far less than 0.1 cubic kilometers in total. And that explains a lot about the last fissures not even being noticed in the general mayhem, and that there is even little evidence remaining of them having taken place.

So, the beginning we know a lot about, but the end is a bit shady.

15 – 29 May: Small but noticeable earthquakes. It is hard to say if these had anything to do with the upcoming Skaftár Fires, or if this was due to the ongoing eruption at Grimsvötn. I have come to believe that they had more to do with the ongoing eruption at Grimsvötn than anything else, but this is not possible to prove even though it is likely.

29 May – 8 June: Strong earthquakes from the area of where the Skaftár Fires would soon start. They were both increasing in numbers and in strength over time. Known damage to houses and witness accounts give at hand that many earthquakes where 5M or above in strength. The residents in the vicinity had to move out into tents to not have their houses fall in over them.

8 June: Large scale explosions and phreatic ash fall over Sída. The start of the Skaftár Fires.

9 June 1783 – 7 February 1784: Fissures 1 to 10 (with most likely two more sub glacial fissures) opened up along a line of 27 kilometers. During the eruption several large scale lava surges came pouring out of what used to be river gorges. During the highest rates of effusion entire rivers stopped flowing.

Summer and autumn 1783: Large scale ash fall over all of Europe, Canada, Greenland and Spitsbergen. Ash from this period is found in drill cores from all over the northern hemisphere.

24 November 1783: A very large earthquake ranging from 6 to 7 M happened in the Skaftár district. This was probably caused as an entire section of the fault line collapsed.

The East Northeastern Fires

Contemporary witnesses talked about a second set of Fires during the Skaftár Fires eruption. This has been discredited by later researchers as a geographical mistake by the witnesses. More or less they have stated that the local witnesses did not know how the land around them looked like and that they did not know what part of Iceland was where.

Photograph of Sidujökull, Geirvörtur to the left and Kerlingarfjöll at far right. Image taken by Aggi och Kristrun. Their photographs can be found at

In the end it seems like they knew exactly what they were talking about. During a campaign in 2010 samples were taken and analyzed from Eldgigur, Hágöngur (rubble) and East Geirvörtur that were found to be showing consistencies with general Lakí lavas. Additional testing places the samples close to the given time frame. The area of these samples are following the eastern side of Sídujökull and is situated on the eastern side of the Grimsvötn fissure swarm, whereas the Skaftár Fires took place at the western side. Since no other eruptions have been noted in the area either before or after the Skaftár Fires it seems like the old Priest knew exactly what he was talking about.

There is also reason to believe that this secondary fissure line extended via the known central volcanoes on the Grimsvötn fissure swarm up to Thordharhyrna which is the point that the witnesses said erupted.

Grimsvötn eruptions

Not only did Grimsvötn erupt prior to the Skaftár Fires, it also erupted at least 3 times during the Skaftár Fires. One of the eruptions that have previously been attributed to Grimsvötn was most likely Thordharhyrna.

After the Skaftár Fires ended Grimsvötn continued to erupt until April 1784. Grimsvötn once again had a final eruption April 4 – 26 1785. After this final eruption Grimsvötn had a 38 year long rest, probably one of the longest uneruptive periods in the last 500 years.

In the next installment I will get down to the nitty gritty details of the ash and dust of the Lakí eruption.


20 thoughts on “Laki deconstructed IV. A Timeline for Destruction

  1. A question… if you answered it in your article, I apologize in advance. How big (volume) was Skaftár Fires. Is this the same eruption that had Europe choking on gas, as well?

    • Yes. It is in the earlier instalments: this is the Icelandic name for the Laki eruption.

    • As an addendum. “Skaftár Fires” is actually a better name since Laki itself did not erupt. The fissure ran up to the base of Laki, then continued onward on the other side of it. So, it had fissure eruptions occuring on either side, but none from it’s edifice. But… Laki is what it is known by in most of the literature.

      Related, but unrelated side note. Some time back I did a bit of reverse engineering on the reported effects of a few quakes leading to the 1783 event. Working though some of the USGS formulas I came up with the “unzipping” as being a sequence of roughly magnitude 5.0 events.

      (I took the reported effects and equated that to a specific MMI level, then worked out what size quake at the fissure’s distance would be needed to cause the observed effect. This is a highly dubious method since it assumed the related quake was at the fissure.) So take my Mag 5.0 statement with a grain of salt. It could be completely wrong.

      As for the equations I used… they came from the USGS “Did You Feel It” paper by Atkinson and Wald. Seismological Research Letters 78(3):362-368 · May 2007

  2. Is it possible there’s a mistake in the timeframe?
    It jumps from “9 June – 7 February 1784 (…)” to “Summer and autumn 1783 (…)”. Maybe I’m overlooking something. Looking forward to the next installment!

  3. M3 at the summit of Mauna Loa. Not seen on HVO as their monitors remain frozen.

  4. Laki flowed at maximum in the lava flow channels an estimated
    Around 6000 cubic meters a second!

    Thats freaking 60 times faster than Fissure 8 s lava flow

    Pretty much all of Skaftareldahraun is rough Aá lava flow suggesting very fast emplacement.
    The lava channels are beautyful too.

    Laki was very fluid and quickly moving in its lava channels with large standing waves and lava torrents and eddies
    And large boulder slabs of channel edges quickly moving in the rivers.

    Eyewittness tells of a very fast moving ”river of fire flowing like a spring flood”

    • Laki was a mini flood basalt eruption. It is interesting to see in what way it was similar and in what different from the run-of-the-mill rift eruptions such as Holuhraun and Leilani. When Carl wrote this, neither of those had happened. We had had Krafla, but Carl makes the point how different that was from Laki. As for Holuhraun, Laki erupted more in two days than Holuhraun did in 6 months.

  5. Looks like Grímsvötn approves of this article. This pair of quakes made the CSM graph step up a notch:

    10.05.2020 01:45:45 64.483 -17.499 5.3 km 2.3 99.0 13.9 km NW of Grímsfjall
    10.05.2020 00:53:16 64.417 -17.246 1.7 km 2.2 99.0 1.9 km NE of Grímsfjall

  6. 4 quakes on multiple of 15 minute increments. looks like one big quake. somebody probably has a theory…???

    • Large earthquakes ~10 km deep under Loihi can reach M4-M 5, swarms seem to repeat from time to time at that same depth, I am not sure what is the cause but I think they usually follow a mainshock-aftershock distribution. While the source of strain is probably volcanic the earthquake swarms may rather controlled by tectonic cycles, but it could also indicate pressure build-up, hard to know.

      If an eruption happens I would expect to see shallower seismicity but the only way to confirm an eruption would be if there is some large enough change in the shape of the volcano that is later detected by a bathymetry survey, like the 1996 eruption which collapsed the summit.

        • A high frequency (spasmodic) tremor detected as far as Hualalai, could be either from the Pahala area or Loihi.

          • HVO now locates the event as a typical deep tremor south of Pahala:

            M 1.9 – 18km SSE of Pahala, Hawaii – 42.8 km depth – 2020-05-12 09:31:07 (UTC)

  7. I just read a report on Lohi and the enormous amount of quakes it had in a month back in 1996.

    Earthquakes and Noise
    Between July 16 and August 9,1996, over
    4000 Loihi earthquakes were detected by the
    seismic network of the U. S. Geological Survey’s Hawaiian Volcano Observatory (HVO).
    The initial phase of activity, consisting of 72
    For more information, contact Fred K. Duennebier, Dept. of Geology and Geophysics,
    School of Ocean and Earth Sciences and
    Technology, University of Hawaii, Honolulu,
    HI 96822.
    located earthquakes, continued for two days.
    After 30 hours of quiescence, activity resumed and continued at a higher rate, averaging over 88 located events per day for the
    next 10 days before slowing.
    Preliminary locations calculated using
    the HVO seismic network data placed the majority of events between depths of 10 and 14
    km shallowing seaward. However, P wave arrivals at an OBS deployed on Loihi summit
    during the third week of the swarm arrive
    about two seconds early, suggesting that the
    velocity model used for the island of Hawaii
    is inappropriate beneath Loihi and that initial
    locations are inaccurate. Using HVO’s preferred velocity model under Hawaii and
    lower velocities at shallow depths under
    Loihi, reasonable hypocentral locations are
    obtained between 7 and 8 km depth (Figure
    1). Despite the obvious topographic modifications of the summit, few shallow earthquakes
    (between 0 and 5 km) were located.
    Sonobuoys dropped from the K-O-K to listen for earthquakes and eruption sounds detected bangs, pops, and grinding noises with
    frequencies from tens to several hundred
    hertz at three distinct locations on the northeast side of the summit, moving from south to
    north with time. An active area just north of
    East Pit was detected during the RRC, and two
    areas were located during the first LONO, one
    on the east flank at a depth of about 1600 m
    and one just north of the summit. Attention to
    other priorities precluded detailed surveys of
    these sites, but turbid water was observed
    drifting in from these areas during submersible dives.

    • Interesting, so I guess we may not see shallow seismicity even if it erupted. The recent swarming now looks more like a volcanic-tectonic sequence and if the earthquakes are poorly located in depth then there some chance this could be a dike intrusion or eruption.

Comments are closed.