I have lately read a lot of comments about Iceland being boring and calm. In reality nothing could be more wrong, Iceland is brimming with activity. So, let us take a quick look at some interesting volcanoes before we go to the namesake of this article.
Volcanic activity in general
If we start in the far north there has been a persistant swarm with a few M3+ earthquakes located at a volcano to the east of Grímsey. The volcano in question is badly studied, but it is known to contain at least one highly active hydrothermal field. If one play around a bit with the earthquakes larger than M1.5 a pattern emerges that form a stack rising from 14km depth up to around 10 km depth. This is normally a sign of rising magma in Iceland. It is though a good idea to remember that this area is well known for its numerous intrusions and intense episodes of seismic unrest, so this may lead to nothing. Or, if we are really lucky it could lead to an eruption with an island formation.
Bárðarbungas seismic unrest is slowly abating after the Holuhraun eruption. In the aftermath the level of seismic activity has been one of the highest recorded in human history. The reason for this is that the bottom of the caldera has continued to subside as a whole plug downwards due to loss of pressure during the eruption. This function has been the primary source for the earthquakes. After all, the caldera “plug” is basically a stone that is 339 cubic kilometer large and weighs 950 000 000 000 tons. Dropping a stone that size about 100 meters is indeed a noisy business.
The loss of pressure has also caused deeper and smaller earthquakes that are signs of fresh magma rising up from the depth and confusingly enough these earthquakes are caused by increased pressure. For now these earthquakes will not cause an eruption since it will take quite some time to rebuild enough pressure for an eruption to be possible.
In the end Bárðarbunga is far less interesting than it seems for a casual observer. So, without much further a due let us leave this particular giant and move on south.
Lately there has been a lot of talk about Katla and the ongoing seismic unrest. Only problem is that the bulk of the earthquakes belong to the Gódabunga system and not to Katla in and of itself. Gódabunga is a nascent magma reservoir emplaced at 4km depth and may form a future volcanic centre in between Katla and Eyjafjallajökull. The seismic unrest has been shallow and may have caused an increase in hydrothermal activity. Nothing points towards an upcoming eruption at Katla at this point, but that might change with time.
Now it is time to turn to the two likely volcanoes to erupt soon in Iceland.
The highly enigmatic volcano of Hekla is known to be a very quiet volcano between eruptions with perhaps one or two recorded earthquakes every year between eruptions.
As an eruption starts you get between 90 minutes and 32 minutes of minor earthquake activity prior to the highly explosive onset of eruption.
During the last 5 months there have been weekly earthquakes and sometimes even daily earthquakes directly under the volcano ranging from 7km depth to 0.1km depth. The magnitude of the earthquakes has been minor ranging between M0.1 to M1.5.
To be quite clear, no period like this has ever been recorded at Hekla. But, before you go off screaming wolf there is a factor we need to look at and that is that the seismograph network around Hekla has been extremely upgraded in the years after the last eruption in 2000. The individual seismographs are now between 10 and 100 times more sensitive, and there are quite a lot more of them now. If you look at the system as a whole it is about 1 000 times more sensitive than before the last eruption. So, it may be that Hekla has had episodes like this before the last eruptions and that we could not see it. This is called technological skew and it can cause over-interpretation of data.
But, if we look at the data that we do have in the last few years since the last large upgrade we see that there has been no such episode. The million dollar question is if Hekla will erupt soon. I would say that it might do that, but that we will not know until a few minutes before the eruption starts when a brief smattering of between 10 and 50 earthquakes happens in a very short time frame. Let us just say that Hekla is not a good spot to hike on right now.
If Hekla is enigmatic and unpredictable Grímsvötn is the opposite. Prior to the 2011 eruption I developed a predictive model for Grímsvötn that accurately predicted the upcoming eruption 6 months before it actually happened, as such it is probably the earliest predicted volcanic eruption that actually happened when it was predicted to happen.
If you have a fine enough set of data you could for all point and purposes exactly predict any volcano like Grímsvötn years in advance using finite element breaking point threshold analysis. Sadly there is no such data on Grímsvötn. But, we roughly know the location and distribution of the magma reservoirs under the volcano, and there is a bit of data from other large caldera volcanoes in Iceland that I used as approximations for Grímsvötn.
Another thing we know is that Grímsvötn is a very steady erupter and that is pointing to the volcano having a fairly steady influx of magma from depth via open conduits. The open conduits are deduced from the fact that we do not see earthquake stacks forming as the magma move upwards into the upper magma reservoirs.
So, if magma influx is a steady number we should get an inverted logarithmic build up of the strain. And that strain build up should be evidenced in a logarithmic curve of cumulative seismic moment.
And to finish the predictive model off we find that the finite element modeling points towards there being a fairly stable point where one of the reservoirs ruptures.
Now I am done with the highly simplified technical mumbojumbo and we are ready to go on to actual data. But, before that I need to point out that to do modeling like this you need to rely on a lot of data that requires a tremendous amount of research that was done by a lot of good scientists at the University of Iceland and the Icelandic Met Office. Secondly you need data from quite a few eruptions to check your model against, and it is pretty much only Grimsvötn that this is possible at due to the good work of the University of Iceland and the Icelandic Met Office.
Now time to leave the mathematical modeling and look at real world data. In that real world data we indeed find a breaking point where the pressure exceeds what a reservoir at Grímsvötn can hold, and it is possible to easily follow that on a graph representing the energy released from Grímsvötn earthquakes equal and larger than M1.
The graph below shows curves with starting points set to the day after the previous eruption and that run up to the moment the eruption starts. Then the graph is reset. For some reason this graph only show the two previous eruptions and the current graph running up towards the next eruption. It is though good to know that the time between the 1996 Gjálp eruption and the 1998 eruption looked pretty much the same as the other two with the 1998 eruption breaking point being slap bang in between the 2004 and 2011 breaking points.
Here is a neat thing. You do not need a higher degree in physics and pages of weird mathematic symbols to predict Grimsvötn like I did. I just did it to prove a theory, in the real world you just need to keep an eye on the graph and watch what the cumulative seismic moment is and you can predict the time of eruption as well as I can.
Currently a temporal volcanic forecast for Grimsvötn would say that it will erupt at any time between 100 days and 3 years depending on the rate of cumulative seismic moment increase, and that as we draw closer that prediction would grow far more exact. If you wish you can do statistical modeling over what curve would be most likely and you would come up with 2 years as the most likely time for the next eruption unless something changes in the system.
Another thing that the finite element theory tells us is that the eruptive loci will change from eruption to eruption since there is more than one magma reservoir. The logic behind this is that the pressure will always be to lower in the reservoir that just erupted compared to the others.
1996 the eruption was a rift eruption emanating out of the northern reservoir, in 1998 the eruption came from the southern caldera wall, in 2004 the eruption came from the eastern side of the caldera; in 2011 it was once more from the southern side of the caldera.
If we now look at the location of the earthquakes that has happened in the last six months we find that the earthquakes once more follow the predictive modeling with the second largest release of seismic energy being towards the Gjálp fissure and the largest being towards the east with very little energy being released at the southern caldera wall.
So, we can even limit the location of the upcoming eruption to either the NE caldera wall or the E caldera wall. The eastern earthquakes have been forming lineaments due east towards Hamarínn Volcano indicating that the next eruption may be a radial fissure eruption like the 1996 Gjálp eruption. The large jump in the Cumulative Seismic Moment graph was a M3+ earthquake that happened 1/3rd of the way towards Hamarínn smack bang on the indicated radial fissure and could indicate a future eruption site.
So, as you can see forecasting volcanoes does not involve watching the moon, reading tarot-cards, tea-leaves or raccoon gizzards. Instead it involves logic, research, mathematical modeling, nice visual graphs and foremost a lot of empirical data. Then you test your model, improve it with new data and see if it works once more.
Before I leave you to watch a very slowly changing graph at the excellent IMO page called Vöktun Vatnajökuls I want to once more say that this particular way of predicting an eruption only works for Grimsvötn so far. It may in the end be possible to extend it to all Icelandic main type volcanoes, but that will take a lot of time and way more data than we currently have at this point. And before anyone ask. I used a completely different model to predict Bárðarbungas 2014-2015 eruption, but I will get back to how on earth I pulled that one off sometime in the future (hint, it did not involve reading badger gizzards).
As some have noticed Grimsvötn has suffered from a minor Jökulhlaup in the last few days. That one may have been caused by what are 3 episodes of hydrothermal unrest. Hydrothermal activity in Grimsvötn is known to increase as we near an eruption.