I got the honour and privilege to be the author of the first blog post of 2018. So let me take this opportunity to wish all our readers, visitors and the managing team a very happy and healthy new year.
The Bardarbunga eruption in 2014 was impressive in many ways. It was a rifting event, it took place more than 40km away from the magma source, it jumped fissure swarms, it was the largest eruption by mass and volume in Iceland in the past 230 years, it erupted ~1,6km3 of lava over ~85km2 area and it is responsible for one of the best recorded cases of caldera collapse in the world, and the first in Iceland since the Askja collapse in 1875. The gradual caldera collapse of Bardarbunga was very energetic, with many earthquakes above M4/M5 on the caldera ring faults, caused by the centre floor dropping, as the magma was transported out of the magma chamber under it, and lasted for the duration of the eruption.
In the end, the caldera floor dropped by 65m at the lowest point and over ~110km2 area in total (area of minimum 1m subsidence). But just a few months after the eruption, when all seemed to quiet down, the caldera ring faults have reactivated with a growing number of M3+ earthquakes, that continue till this day.
In this rather short installment, I would like to shed some light on the post-eruptive seismic patterns of Bardarbunga, compared to the caldera collapse seismicity during the 2014-2015 eruption, while keeping it as simple as possible with mostly my own graphical presentations. The data I am using is from the manually checked/corrected earthquake catalogue from IMO Iceland.
The intention of this post is not to make theories or over-assumptions, but just to present the data as objectively as possible. I am not a professional volcanologist/seismologist, but just an amateur/enthusiast, with the passion for volcanoes and geology, more specialised in producing plots/graphics rather than making deep theories and thesis’s.
There are 30+ graphics in this post (28 + 1 video of my own work), so give it some time to load everything.
Now that all the disclaimers are out of the way, we can begin. 🙂
2014-2015 Caldera collapse
On August 16th, an earthquake swarm began between Bardarbunga and Kistufell. On 17th and 18th August, the swarm focused more on the area around Bardarbunga, specifically towards the SE end of the caldera, from where a lateral dike has formed towards SE. After going 10km SE it made a 90° turn towards NE, and by August 20th, the dike path was already 30km long, with an estimated volume of around 0.25km3.
The magma that was already in the dike and was on its way towards NE, most likely came from the magma reservoir under Bardarbunga. Judging by the seismicity signature, the magma source for the forming lateral dike was between 10-14km depth, as seen on the cross section below. Under B there is the likely inward dipping caldera ring fault, with a lateral extension of earthquakes between 10-14km depth, where the likely magma outflow path was. The direction and shape of the seismicity corresponds to a transport dike, with the seismic “tail” being only 0,5-1km wide, when going into the turning point. There is a possibility that some of the magma was also transported from a deeper source.
With the onset of the seismic swarm and the formation of the lateral dike out of the magma chamber on August 16th, deflation was also being recorded on the same day, as a response to the magma chamber being drained. But the caldera ring fault seismicity was low from August 16th and 19th, despite over 0.2km3 of magma already being drained. The reason was likely the initial elasticity of the intra-caldera floor/magma chamber roof, from here on being simply referred to as “the plug”. Another indicator of the elasticity of the plug, was the fact that it had a smooth deformation surface, meaning that it did not develop any visible concentric ring faults, which would also appear on the ice surface, and would also have a seismic trail at the surface. There was though seismicity within the plug itself as the stress fields were highly altered. Smooth deformation was also likely due to the high aspect ratio of the plug, being taller than wider (6-8km wide and 10-11km tall). Such an aspect ratio of the caldera plug, provides a better internal stability (support) and can withstand more deformation before developing bigger internal cracks/concentric faults. That is compared to a very low aspect ratio caldera plug/floor, like at Grimsvotn (4-6km wide and 1-2km tall), which is subjected to many cycles of rapid inflation/deflation, and can become a major problem down the line, when once such cycle of stress will exceed the structural integrity of the magma chamber roof, and a new larger explosive caldera forming event is possible.
On August 20th, the elasticity limit of the Bardarbunga caldera plug was likely reached and stronger earthquakes began appearing on the north and south ring faults. This was likely the date of the plug drop/slip onset. By August 28th, activity increased, as the full ring fault was fully activated. First to have major earthquakes was the southern rim, which was likely under a bit more stress since the magma began the outflow from the SE margin. The activity soon began on the north rim as well, and was alternating between north and south without any specific pattern, with more energy release on the north rim.
The eruption at Holuhraun began on August 31st.
At this point, the plug drop was acting as a hydraulic piston. It was responding to the reducing pressure and volume of the magma chamber due to the eruption, but at the same time, as it was dropping and pressing onto the magma chamber, it was increasing the pressure in the chamber and the outgoing dike due to compression. With the individual fast drops of 20-80cm, that were accompanied by M5+ earthquakes, measurable pressure waves were sent across the dike from Bardarbunga to the area of the eruption, from the pressure increase as the plug was pressing down. So simultaneously, the eruption was driving the caldera collapse, and the caldera collapse was driving the eruption. A rather rare, but perfect synergy.
On September 13th, a GPS station (BARC) was set up on the caldera floor, which was measuring the descent of the caldera floor, together with aerial radar and satellite measurements. The rate of decent was nearly exponential, and was perfectly correlated with the eruption rate at Holuhraun. This, combined with petrology and seismic analysis, is a solid proof that the magma source of the Holuhraun eruption was from (one of) the Bardarbunga magma reservoirs. The end and start of the caldera collapse was also well correlated with the eruption/magma transport.
The seismicity during eruption was very energetic, and left behind a very distinctive pattern of ring faults. But ring faulting is never just a simple straight line. Nevertheless, two distinct general faults have appeared, on the north and south rim. The magma reservoir under Bardarbunga was modelled to be displaced/expanded a bit towards N/NE at depth, relative to the volcano surface. This is also likely the reason why the north ring fault has a slight outward dip angle (80-85°) and had mainly thrust faulting, while the south ring fault has a near vertical dip (85-90°), and had mostly normal faulting. And it is also likely why the west margin of the caldera had weaker subsidence gradients and less strain build.
The subsidence gradients were also strongest on the north ring fault, as the plug was being pulled N/NE under the rim, which also causes more strain with the surrounding crust. The NE outer rim actually recorded some uplift as the plug was sliding under it, and slightly pushing the overlaying crust upward. While on the south ring fault, the plug was being pulled straight down or slightly away from the rim and the surrounding crust, and had less strain, compared to the north ring fault.
On the next image we can see the approximate angles of the ring faults on the western caldera margin, and a very nice outline of the plug. B is the north rim and E is the south rim. The north rim is much more noisy and has an outward dip. The south plug has an average inward vertical dip, but is a bit more complex down lower. Some intra-plug seismicity is seen, as a response to stress fields within the plug itself.
Next image shows the same field, but only M3+ earthquakes. This outlines the fault orientations a bit better, since it shows the area with most strain build/energy release.
Going over to the eastern margin, things are just as complex. The fault angles are mainly the same, except for the north rim, which perhaps has a bit more outward dip down in the bottom half, especially in the bottom third, where the angle increases even more and where there is coincidentally more strain release. The south rim retains the same angle.
When looking at seismicity, I can’t help not to look for patterns that might suggest where the magma reservoir is. The petrology, deformation and gas analysis for Bardarbunga have all put the magma source with confidence at around 10-16km depth. That is not the actual location of the magma chamber, but just the outer error margins.
Looking at the seismicity from two different perspectives, we can see a slight sign of a seismic discontinuity between 12-15km depth. It is not a seismic free zone, but it is a clear change of pattern between upper and deeper activity, and agrees with the model depths and sizes (ellipsoid-sill-like body). There was continued deeper activity under Bardarbunga during eruption, indicating a possible upward magma transport as the magma was flowing out of the chamber which was causing under-pressure in the chamber (though quickly mitigated by the plug drop).
A cross section on a W-E axis also shows a near seismic free zone around the same depth, and where the outflow conduit was towards E/SE. But magma reservoirs are never of a perfect shape and can be very complex.
Looking only at 13-16km depth, there is a void under the caldera, getting complex towards NE.
The eruption lasted for 181 days. Despite the plug acting as a compressing piston, the overall pressure was reducing as more and more magma was leaving the reservoir. The whole system reached an equilibrium of pressure, and the eruption ended on February 27th 2015, leaving behind a very bright seismic picture.
After the end of the eruption, there was continued low magnitude seismicity in the dike path, as the magma cools and slowly contracts. The seismicity also continued around the caldera, both around and at the ring faults. An M3+ earthquake was more an exception than a rule. But that rule changed in September 2015, when M3+ earthquakes became a regular occurrence. But it went further, with M3+ earthquakes that were getting nearer to the M4 mark. And in 2016, the M4 earthquakes became the new regular periodic occurrence. Till this day (2nd January 2018), the strongest post-eruptive earthquake at Bardarbunga, is an M4.7 on the north ring fault, at around 5km depth. There can be many causes for the increased seismicity, or all combined to some extent. Either the inflation under the volcano pressing upwards, the still active ring faults as the plug is continuing to adjust, general tectonic stresses, interactions with the stress fields of surrounding volcanoes and deformation, etc… The caldera collapse was a violent event, and it takes some time for the ring faults to fully release all the strain, especially since the ground in Iceland is constantly on the move.
North vs south, east vs west….
If we now take a look at the seismic profiles of Bardarbunga post eruption, we can see a similar but not the same picture as during eruption. Below is a vertical cross section of the west margin of the caldera. On the right display you can see how the stronger magnitudes nicely outline the caldera ring faults. I decided to use all magnitudes, to show a much better contrast between the ring faults and surrounding stress fields. On the left, I have outlined some of the vertical trends, mostly fault dip/orientation. It is of course much more complex that a simple straight line. The north rim has retained the same general angle, while the south rim got a bit of a deformed look, with an inward/outward dip complex from 6-10km depth. It almost feels like there would be an external force pressing towards the structure on a SE->NW axis. I have added possible chamber/sill outlines.
Going to central area, things normalize a bit. The north rim has the known fault dip, and the south one is also closer to eruptive look. But there is still the dip anomaly below 6km. Here I have also added a potential sill location/s.
Moving over to the east, this is the region that has retained most of its eruption days dip angles.
Looking at the specific rim fault, they have a generally normal look, with a slight bias towards east with depth. It retains the same general shape and orientation as during the eruptive/collapse phase.
The full profile on the image below, shows the deeper activity, likely magma transport, and we see the upper structure of Bardarbunga. The known magma storage is around 10-15km based on gas, petrology and ground deformation data. The new source of deformation and seismicity could be the same magma storage that provided magma for the 2014/2015 eruption which is getting rejuvenated, or could be a deeper sill storage. The whole deep system of Bardarbunga is not well known/understood. What is certain, is that Bardarbunga sits almost directly on top of the Iceland plume, and it regularly receives fresh material from the mantle, more-so during plume pulse phases.
Now another potential anomaly in all this, is seismicity that is a bit away from the caldera rim, but connected to it on the north side. It is a batch of seismicity towards NW, trending away from the foot of the ring fault. It looks like a continuation of the ring fault, or an associated stress field. But it is not entirely impossible that it is a response to the changes in the stress field due to an inflating body. Its orientation is NW/SE.
The GPS deformation vectors (detrended) do confirm a deeper source of inflation around Bardarbunga. Iceland Met office has also confirmed that inflation has started soon after the eruption, based on all the data they have at their disposal. The closest station (KISA) has recorded around 80-100mm of uplift since the end of eruption. But the area is tightly packed with volcanoes, so the interpretation of deformation should not be linear. There are many potential sources of deformation, together with general plate tectonics.
There is no known (or public) information about the depth of the caldera floor or any potential recovery since eruption. Such accurate observations cost money, which is usually granted when there is a confirmed/realistic risk of an immediate eruption or danger to public safety. Bardarbunga is currently not posing an immediate eruption risk, but it is being closely monitored, since it is showing signs of a steady post-eruption recovery.
Having a quick look at the post eruptive trends at the Bardarbunga caldera. We could see above that the magnitudes have a general positive trend. It goes the same for both the north and south rim. I have added simple linear trend lines on these graphics, just for the sake of better orientation. An interesting point is July 2017. After that, the amount of M3 earthquakes has reduced, but the amount of M4 earthquakes has increased on both rims. This generally means that the earthquake magnitudes and energy release has increased, despite the reduction of M3 earthquakes. Likely some colder/denser/stronger material is being cracked. Future activity will show if this is a new trend or a phase.
Location wise, it is a very interesting story. Both on the north and south rim, earthquakes (M3+) have a slight NE trending. This likely has more to do with general plate tectonics, than anything coming from Bardarbunga itself. The strongest easterly trend is on the north rim. Sometimes this can be used as a “passive GPS” signal, but one has to consider all the tectonic factors that can influence such a signal.
Looking at depth, we had a general trend towards deeper earthquakes. But just like with M3 quakes above, after July 2017 the depths somehow reduced. After a 3-month break, the depths are slowly increasing again. Something has likely changed in July 2017, or some point was reached.
As far as energy release goes, the north rim has a big advantage over the south rim. That is reflective of the caldera collapse phase. The northern rim fault has an outward dip angle, has thrust faulting and so interacts much more with the overlaying crust, building more strain. I used a basic energy release formula for this, and modified it a bit. But the numbers are not as important here, since the focus is mainly on the trend and ratio of the energy release between the north and south ring faults.
Speaking of faults, I mapped approximate outlines of ring fault during and after the eruption. It is based on M3+ seismic data, and is just an estimate, since in nature, faults and especially ring faults can be very complex. So I call this a graphic of “potential” faulting. The collapse faulting had quite a circular shape, outlining the overall rim faulting caused by the caldera collapse. Since it is color coded, it shows the slight outward dip angle on the north rim, while the south rim is more uniform in angle by depth.
The post eruptive faulting is a bit less circular and more disperse. South rim still has a similar angle at the eastern side, but it has anomalies on the western margin, just like I showed on the cross sections higher up in the post, 6km and deeper. The faulting here does not reach so deep, since there was much more motion and strain in the stress field during the eruption and during the caldera collapse.
For the post finish line, the full profile image shows a comparison between eruption and post eruption vertical profile of the caldera, using M3+ events. The most obvious thing is that the southern ring fault has moved north by a full kilometer, which was also partially seen by the latitude and longitude trends on the graphics above.
The best way to slowly end this (now rather long) instalment, is by putting everything into a seismic perspective. I made an HD plot, that shows the seismic past of Bardarbunga, back to 1995, or at least as much as it was recorded in the IMO earthquake catalogue. There is a general lack of weaker magnitude earthquakes up to around 2011. That is generally known as a “technological skew”, a term I borrowed from Carl. It basically means that the apparent lack of low magnitude earthquakes prior to 2011 is an artificial signal, because the seismograph network was not dense/sensitive enough to record lower magnitude events compared to recent years.
Nicely seen is the period of frequent deeper intrusions into the system between 2006 and 2011. It is reaching down to around 25km sharp. Why such a discontinuity at 25km depth? It is possible that the feeder system up to that point is semi-open, meaning that magma can perhaps creep upwards without much seismic noise, through pre-existing dikes/paths. It can also mean that the crust below that depth is hotter and more elastic/ductile, not being able to accumulate a lot of strain.
And this would not be a fully graphical post without a video/animation. I have made a simple spin-around video of the Bardarbunga seismic profile, using post-eruption data. The M3+ earthquakes are magnified and have green-reddish colour gradients by depth, which shows the north-south ring faults.
Bardarbunga will eventually erupt again one day. That is (almost) a mathematical certainty (unless the magma source somehow cuts off). The fact is that the 2014/2015 eruption caused a few changes in the system and in the stress fields around the volcano. It will take many decades for Bardarbunga to replace all the lost magma volume with fresh material. But it will take much less time to get the magma reservoir pressure back to pre-eruption levels, thanks to the caldera collapse, and the plug which somewhat mitigated the post eruptive under-pressure in the magma reservoir. This means that once the reservoir goes into over-pressure, magma will likely try to find a new path out of the reservoir. Basalt magma is known to be hot and generally “liquid”, which gives it good flowing properties. It means that it is easier for basalt to find a weak spot and to crack out of the magma reservoir in a lateral (or any) fashion, rather than erupting vertically from a ring fault. At the end, just like water or air, basaltic magma will flow down (or up) the road of least resistance.
This gives two possible scenarios for future eruptions.
-One, is another lateral dike event, similar to the last event, with unpredictable development, since it depends on the size, depth and the direction of the lateral dike.
-Second, is the possibility of an eruption upwards through the ring fault. There was a lot of motion and cracking and energy release on the both ring faults, north and south. There is a chance that once the magma reservoir is over-pressurised, magma can now move upwards if it finds or cracks a way. This largely depends on the strength of the host rock around the magma reservoir, compared to the integrity of the ring fault on the bottom. Both faults are much more fractured than before the 2014/2015 eruption, but at that point the magma was already flowing out of the chamber via the lateral dike/outflow, and had no chance of going upwards, while the lateral outflow was open.
Another thing that favours the ring fault eruption is the faulting. If i had to chose which ring fault could transport magma upwards easier, i would go for the south one, no questions asked, and I will explain why…
Looking at faulting mechanisms, we see that the north rim has reverse faulting and was pulled under the surrounding crust, heavily interacting with it. That is why it released much more energy, since there was also much more strain build, because it was under compression, while pushing/pulling the plug wall and crust wall together under a steep angle, really grinding each other away.
The south rim on the other hand, has a slight inward to vertical dip angle, and was sliding down and away from the surrounding crust, which was causing extension. There is still wall-crust interaction, but since there is normal faulting and extension, there is not as much strain build as on the north side where there was more direct and violent interaction between the plug and the crust.
This leads me to believe that it would be much easier for magma to creep upwards through the south rim, since as it is under extension, it would be easier for magma to find openings or to make its own path/cracks, opposite to the north part which is being pressed together. It is still possible for magma to move up on the north side, but it would take more effort. There are cauldrons in the ice on the south and east side of Bardarbunga, and even steam coming out of the cauldrons. That is also an evidence that the rims are geothermically active, with likely magma and heat transport.
An eruption through the ring fault would most certainly be explosive, at least at first, as the magma would intersect ice, which is always an explosive combination.
Bardarbunga is tho more known for its massive fissure eruptions, like 8.600 years ago, when it produced the largest flood basalt since the end of the last ice age, with lava flows going over 100km, all the way to the south coast.
Just before the settlement of Iceland, Bardarbunga created a dike, over 50km long, towards SW, that reached all the way to Torfajökull volcano in 871 A.D. It caused an explosive eruption, sending ash across Iceland. There was enough ash, that it left a distinctive mark in the soil, known as the “settlement layer”.
In 1477, Bardarbunga did it again, this time with another eruption from the Veiðivötn fissure swarm.
Another big flood basalt is always an option, but at this point it is not very likely. Though with another plume pulse arriving, or already beginning, one can never fully dismiss such an option somewhere down the line.
This brings us to the end of the first post of 2018. I hope and know, that there will be lots more to come over the year, and for many years to come….
Andrej Flis, (Down Under @Recretos)