Grimsvotn (‘Grim’s lakes’) is Iceland’s secret. Of all its volcanoes, this is the most frequent erupter, exploding as often as every 5-10 year. It also causes jokulhaups with decadal frequency. And worst of all, it is a mass murderer, which has killed as many as a quarter of Iceland’s population. And all of this is done from its hiding place, like a troll living underneath the immense ice sheet of Vatnajokull.
There are three, overlapping calderas. This fact alone already shows its potential: Grimsvotn does not do one-off eruptions. The central caldera is the youngest, and is the location of the most recent eruptions. But the hole of the caldera doesn’t show on the surface, at least when not erupting. Everything is deeply buried under its blanket of ice.
Grimsvotn is the source of the Laki eruption, 1783-1785, which devastated Iceland, caused major casualties in Europe from breathing sulphurous gases, and caused a famine that lead to the French revolution. In Iceland this international diplomatic incident is known as the Skaftar Fires, after the river valley through which the lava flowed. Laki erupted a staggering 15 km3 of lava along a fissure, far from the glacier: Grimsvotn acted its vengeance from a distance. But it wasn’t completely hidden. During Skaftar, there were reports of fires seen from the area of the glacier. Looking at where those reports were made, and which direction they gave, allows the source of those fires to be triangulated. This leaves no doubt: during the Skaftar fires, Grimsvotn also erupted. And if that wasn’t enough evidence, the Laki fissure points directly at Grimsvotn. It did not hide its tracks well enough.
The fissure system associated with Grimsvotn extends for over 100 kilometer. Since the end of the ice age, Grimsvotn has produced some 50 km3of lava, a third of which erupted during Laki. About 20 km3 was erupted outside of the glacier (dominated by Laki), and the rest closer to the volcano. How much came from the caldera(s) itself is not known.
I have argued elsewhere, regarding Elgja, that fissure eruptions are driven by gravity: the weight of the volcano pushes the magma out of the chamber. This means that the eruption must occur at a (much) lower elevation than the peak of the mountain, that the caldera which forms from the emptying magma chamber should have a volume comparable to the magma moved out of the chamber, and that the caldera cannot collapse any lower than the exit point. The bottom of Katla is indeed around the same elevation as its main fissure eruption (Eldgja) and it has roughly to right volume. The youngest (currently active) of the three calderas of Grimsvotn is about 20 km2 in area, and 300 meter deep. That means its volume is too small compared to Laki by a factor of 3. However, the caldera may have been much deeper after that eruption. The caldera is at an elevation of 1700 meter, and Laki is at 800 meter. So the caldera could have been as deep as 900 meter, which gives a volume of 18 km3, very close to the volume erupted by Laki. So I think this caldera may have formed during Laki. It is very young indeed.
The magma chamber would have been thoroughly emptied during the Skaftar fires. Grimsvotn has two magma chambers, one 10 kilometer deep and the other one shallow, between 1 and 5 kilometer according to different studies. Over time, the caldera became shallower again (speculation alert), a combination of rebound and refilling of the upper magma chamber.
Grimsvotn has had over 30 eruptions since 1589, on average 10-15 years apart. Tephra deposits show an average rate of 7 eruptions per century during the past 5000 years, with no obvious quiet periods. But after the Skaftar fires, it remained silent for 38 years.
More recently, Grimsvotn has gone through times of (relative) quiet and times of frequent eruptions. There were eruptions in 1867, 1873, 1883, 1897, 1903, 1922, 1934, 1983, 1998, 2004, and 2011, showing a cluster in late 1800’s, and from 1983. The 1873 and 2011 eruptions were large. Eruptions last around 1 week, and they are explosive events from within the caldera, very different from the Laki fissure eruption. The rising magma interacts with the water underneath the ice, which has collected in a subglacial lake within the caldera. (After an eruption punches through the glacier, the lake becomes briefly visible.) The recent eruptions have all come from the southern edge of the central caldera. Eruptive volumes (tephra) range from 0.1km3 in 1998 to 0.7 km3 in 2011.
An eruption in September 1996 is not in this list. This was a fissure event from Gjalp. There is some uncertainty where the lava came from. The composition of the lava is closer to that of Grimsvotn, but the earthquakes that preceded the eruption were at Bardarbunga. Given what we now know about the reach of Bardarbunga, it is quite possible the eruption came from there, and not from Grimsvotn.
The regular eruptions should make Grimsvotn one of the most predictable volcanoes. Indeed, during the last two cycles, leading to the 2004 and 2011 eruptions, the behaviour developed very similar, with increasing earthquake activity, and horizontal and vertical motion of the nearest GPS (GFUM, on the caldera rim), until a breaking point was reached. The horizontal motion needs to be corrected for the normal continental plate motion: once this is done, it shows additional motion in the years before an eruption, a definite sign of an inflating magma chamber. Vertical motion shows a pattern of a sharp deflation during the eruption, with a brief exponential recovery afterwards. Once this is done, there is linear inflation, with changes to a faster increase. The linear increase is the steady-state influx of magma into the lower magma chamber, and the faster increase is when the shallow magma chamber begins to refill. Earthquake activity also follows this pattern.
At the moment, earthquake activity at Grimsvotn is on the rise but not as fast as during the previous cycles at this point in time, and GFUM is moving up by 3 cm per year (note that only GPS data from July-September is usable. At other times you get huge deviations from winter weather, including snow on the site.)
The seismic moment (earthquake activity added up over time) is also increasing. But compared to the previous cycles, it is well behind schedule. The current cycle included one unusually large earthquake (unusual for Grimsvotn, not for Iceland in general) and perhaps this one should be discounted as not part of the pattern: it may have released unrelated tectonic stress.
Now let’s see whether we can use the earthquakes to predict something about when the next eruption may occur. I use the standard equation for cumulative damage:
where tc is the time when things break. A similar equation is used by Got et al, 2014: An analysis of the nonlinear magma-edifice coupling at Grimsvötn volcano . (They use it for the number of earthquakes rather than for the earthquake moment).
The plot shows the results obtained with this equation. Green is for data leading up to the 2004 eruption, blue the 2011 eruption, and red for the current cycle. The data has been read off from the IMO plots and there may be some inaccuracies. I first fitted the equation to the data leading up to the 2004 eruption. The curve fits the data well if I take the numbers tc=2200 days and k=1.2. When I fit the data leading up to the 2011 eruption in the same way, it yields tc=2550 days, and the same value for k. Interesting, the curve suggests that the 2011 eruption happened slightly ahead of schedule, by some 100 days.
We can also now fit the current cycle in the same way. There are two problems with this: the cycle isn’t far enough yet to give a unique solution, and the fit is quite dependent on the single earthquake around day 1850. Regarding the first problem, the range of dates for which the fit is made to work is quite small. That makes is susceptible to noise, random fluctuations in the earthquakes. I went for a fit which works best towards the later dates, where the random noise becomes (hopefully) less important. For the second problem, I have plotted the cycle both with this jump, and without it (the lower red line at the end). Using the same value of k as found in the previous two cycles, that gives me two possible values for tc (noting that both are still quite uncertain!), tc=3200 days including the jump, and tc=4000 days excluding it. Say 8.5 to 11 years. But be aware that the current cycle is still in an early phase and the fit could easily change – it is not yet well determined.
I can take the outcome as the possible range of dates for the time of the next eruption. Starting from June 2011, the early date corresponds to January 2020, and the later date is June 2022. There is a good chance the eruption may occur between these dates.
As a reality check, how does this compare to the past? In the eruption sequence starting in 1867, you can see that the times between eruptions get longer, and that the time after the 1873 eruption, similar in size to the 2011 one, was 10 years. I find the same lengthening of eruption intervals for the current cycle, and a similar time for the next eruption after a large bang. So there is some kind of agreement with the (very limited) historical data.
I should warn first that making predictions is dangerous, very uncertain, and should strictly be limited to the past. I am failing here on all counts. Grimsvotn is by no means bound by my numbers and will do whatever it wants. Second, it is too early in the sequence to make strong predictions. A small fluctuation in earthquake activity can still change the fit by quite a lot. It would be better to wait until the earthquake activity begins to accelerate, which hasn’t happened yet. If the fit was repeated in 6 months time, the numbers could easily come out quite different.
A final point to make is that after the 2011 eruption, there was a two-year period of very little activity, much more so than after the two preceding eruptions. In fact, the activity over this period is well below what is predicted by the equation. If I leave the first 600 days of the current cycle out, i.e. shift the red curve to left by 600 days, it fits rather better with the blue and green curves. That would still give an eruption around the year 2020.
So this is it! Grimsvotn is a prodigious erupter, varying in style between effusive long-distance fissures, and explosive events within the caldera. The frequency of eruptions is high enough that it is worth trying out predictions – the predictions can be tested against reality within a reasonable amount of time. And although the amount of ash and tephra is not huge, it is significant, and problematic for people and for aircraft. Thus, predicting its future is of economic (and human) value.
Will this prediction work? Wait and see! It assumes that Grimsvotn will continue to behave as it has done in the recent past, and that is far from certain. It ignores the silent two years after 2011. And if the next event is a fissure eruption, all bets are off.
Otherwise, Grimsvotn 2021? You heard it here first.
Albert, August 2017
The contrarian view
A week ago, I suggested to Albert that we should write this article together. Both of us looking at the same data, and both of us trying to predict when the next Grimsvötn eruption is likely to occur. It was meant that I should give the contrarian view to Alberts part.
There are two things in Alberts part that I wish to contraire upon before I get back to contraire the likely date of the next Grimsvötn eruption.
The first part is about the calderas of Grimsvötn. They are not as young as Albert suggests in the article, instead they are likely to have formed during 3 out of the 5 Saksunarvatn tephra’s, something I wrote about in a previous article about Grimsvötn.
The other part is about the source of the Lakí lavas. In a six-part article series I wrote about the chemical composition of the lavas where I found that the lavas are not matching the known samples from the South-Central magma chamber, nor the North-West magma chamber (responsible for the Gjálp 1996 magma).
Grimsvötn is a bad bet for being the culprit for the Skaftár Fires since there are two large central volcanoes SSW along the fissure swarms. The northerly of those are the Háabunga volcano that may, or may not, be a fourth magma chamber of Grimsvötn. The southerly of them is the powerful Þórðarhyrna volcano.
The Þórðarhyrna lavas are closer in composition to the Skaftár Fires lava, but it is not matching well enough. In the previous articles I surmised that the lava originated directly from a deep common magma reservoir that feeds all of the above mentioned central volcanoes.
The reason for that is that I found commonalities pointing towards the magma originating from a deep common source, and that the magmas evolved due to melt inclusions and varying times in situ at the magma chambers of the various volcanoes. And the Skaftár Fires lava was less evolved than the other by a fair degree.
There are also similarities with MORB-magmas (Mid Ocean Ridge Basalt), and the MORB similarities was higher at the SSW end of the Lakí Fissure compared to the NNE beginning.
I do agree with Albert that most fissure eruptions are gravity driven, but not all. Some are driven by other functions, even though we do not fully understand those functions currently. It may also be that we do not understand all functions of the gravity theory of volcanoes yet, and that Albert will turn out to be more correct than he believed.
But, I will give an example of a volcano that throughout its history has had major fissure eruptions with no possibility of them being gravity driven according to the current theory, and that is Vatnafjöll.
For some reason, it has produced at least 3 eruptions as large as Veiðivötn 1477. Those eruptions occurred at the fissure that is the top of the mountain, and to even confound us more, it has not gone caldera even though it is among the top 5 volcanoes in Iceland if we rate it after volume of lava produced. The same also goes for another of the top 5 lava producers in Iceland, Þeistareykir.
And now that I have contraired on that I will return to the point at hand and see if I will reach to a different conclusion than Albert.
The Cumulative Seismic Moment-plot has over the years proven to be a highly useful plot for estimating when a Grimsvötn eruption is likely to occur. We have data for the CSM for 2011, 2004 and 1998 eruptions. The 1998 data is not shown any longer on the public CSM-plot, but it was pretty much exactly in the middle between 2011 and 2004.
The dataset is not long enough to determine if the CSM-values are true for what the lowest and highest values will be for an eruption at Grimsvötn. One way to think about it is that the high rate of eruptions that we are seeing now are weakening the magma reservoirs, so that the CSM-value will decrease over time for an eruption to be possible to occur.
This line of thinking kind of makes sense compared to the not so small fact that the 2011 eruption had the lowest CSM-value at the start of the eruption, and that it still produced the largest of the 3 eruptions.
1998 had a value of 3.5, 2004 had a value of 4 and 2011 a value of 3. Let us therefore toy with the idea that it is possible that an eruption can occur at as low a CSM-value as 2.5, that would set the next possible eruption closer in time.
Now let us ponder the opposite, we know that the Grimsvötn eruption of 2011 was the most voluminous since the Skaftár Fires in Iceland. It is also the single largest eruption since the large VEI-5 eruption of Cerro Hudson in 1992.
It is probable that the size of the eruption lowered the pressure of the magma reservoir sufficiently that it will take a longer time to refill, and that it therefore may be able to withstand more cumulative seismic release than before. If this line of thinking holds true it could be that Grimsvötn may reach a higher CSM-value at perhaps as much as 4.5.
Here I should point out that the CSM-value of Grimsvötn is very low, the same current known number for Katla is in excess of 140 (if my memory serves), and then we only have a partial record for Katla. This has nothing to do with a potential future eruption size, I just mention it to give a background of how little it would take to set of Grimsvötn in regards of an earthquake swarm.
Now we return to that pesky M3 earthquake that occurred between Grimsvötn and Hamarinn. It offsets the CSM-value with roughly 0.5, so it is pretty significant. On one hand, it is not directly related to the pressure of the magma reservoirs of Grimsvötn. On the other hand, it is pointing towards the possibility of a future Gjálp style eruption in, or around, that location.
With that earthquake counted in we have a CSM-value of 1.5 currently, without it we are currently at 1.
This gives at hand that the earliest possible eruption according to the curves of Albert is somewhere between December 2018 to February 2019 if we assume that the pesky earthquake is related and that the true CSM-value is 1.5 and that an eruption can occur at as low a CSM-value as 2.5.
If we instead go for the latest possible scenario of a eruptive CSM-value of 4.5 and discount that pesky earthquake we end up all the way into the summer of 2022.
Is the CSM lying?
I am here going to postulate that the CSM-value is a false presumption for the upcoming eruption, even though I myself have often used it. So, I am here going to contraire myself for the sake of scientific debate.
As evidenced by the Gjálp eruption, Bárdarbunga has for quite some time exerted a significant amount of pressure on Grimsvötn compressing its deep magma reservoir and upper magma chambers. During the 2014-2015 eruption at Holuhraun this pressure dropped quite a bit as evidenced on the GPS-stations. During the eruption, the inflation of Grimsvötn stood still, and even for a period reversed.
This lowered systemic pressure resulted in a period of less earthquakes in Grimsvötn and hid quite a bit of pressure build-up and inflation since the magma system increased slightly in volume.
The second reason that the CSM-value might be off is that the GPS-system is pointing towards enough influx of magma to reset the volcano back to a value higher than prior to the 2011 eruption. And it is when we merge the GPS-data with the CSM-value that we get interesting tools to forecast the next Grimsvötn eruption.
The GPS-data negates the idea of higher CSM-value and increases the likelihood of a lower CSM-value. I will therefore state that likely CSM-span is between 2.5 and 3.5 with the peak at 3. I will also postulate that there is a risk that a future Gjálp-style eruption can occur between Grimsvötn and Hamarinn. I will therefore leave the pesky earthquake in the plot and count our current true CSM-value as 1.5.
If we know superimpose this on the Albert-curve we reach a probable time span for the next eruption ranging between December 2018 and December 2019. The if here is if the upwards motion of the Albert-curve will go as he has formulated it above, it is though likely that it will do so.
I will though here like to point out once more that the CSM-values at Grimsvötn are very low. In fact, there could at any time come a rogue earthquake swarm, or a single M4 earthquake, that punches the value fast into the eruptive ranges. It is though not as likely judging from previous behaviour of Grimsvötn.
In the end nature is a messy beast, and even the best models can be side-tracked by rogue events.
This week there are the usual five riddles with one picture riddle. As usual only I and Gaz has the answers to the riddles. We may award bonus-points for advanced Volcanocafeishness. Good luck with these brainbusters!
|Name||Last week’s points||Total|
|Chris Cookie Cooke||0||1|
- Warm (Spanish speaking) salty waters from the bull
- The one living saint (in Moulin Rouge!)
- (Roccella Tinctoria covered) Basalt of Anthony and Matthew – Urzelina, answered by Irpsit
- Flowing badly read viral kiwi – Nyamuragira, answered by mjf
- French saint + – Boomerang Seamount, Answered by Thomas A
answered by Tomas A