For every person that is interested in volcanoes and complex geology, Iceland is probably something special. A unique place where the MAR intersects a mantle plume, or vice-versa. A beautiful island filled with powerful volcanoes and other complex volcanic and hydrothermal features. All in one nice little spot.
Of the entire Icelandic volcanic lot, the most interesting volcano to me is Katla. Perhaps because I have seen numerous documentaries about it, heard numerous theories back in the days of the Eyjafjallajökull eruption, or perhaps because of its powerful history and the fact that I am likely to see it erupt in my lifetime. Not to mention that unlike most Icelandic volcanoes, I can actually pronounce this one.
A lot was said and done about Katla on this blog in the past, including a great 3-part series titled “What´s going on at Katla“, written in 2012 by Henrik. It covers the basic and important information about Katla, its history and its past and recent activity.
Another great post on Katla was written by Carl, and is titled “Katla and the art of reading data“. It contains a lot of important information which go in-hand with my research.
It is no secret that I know my way around plotting earthquakes or other geological data, in various dimensions. I have learned it myself a few years earlier when I was plotting meteorological data from various models and observations. As my interest in volcanoes and geology grew with time, I decided to try to use my techniques from the weather world above, and apply it to the dark abyss underground.
All the earthquake data that I use is from the Iceland Met Office earthquake catalogue dataset, and contains only manually verified earthquake data (99% quality).
It is very important to note, that while the data is very accurate, there are still some uncertainties in the dataset. Horizontal locations (epicenter) are very good, especially for M1+ quakes, but they can still have a possible minimum margin of error of about 100-200m in all horizontal directions.
The depth (hypocenter) however, is harder to accurately determine. And the accuracy depends first and foremost on the density of the monitoring network and the distance to the closest seismometer. It also depends on the depth (the deeper we go, the harder it is to accurately locate the hypocenetre of the earthquake), and the magnitude of the earthquake, since stronger earthquakes have stronger signals and are easier to accurately locate. Therefor we have to assume an average minimum margin depth error of 500m-1km.
For in-depth scientific research or tomography models, it is suggested and required to deposit a denser network of seismometers, which enables more accurate location of earthquakes. There are also several methods of “earthquake relocation”, which is essentially an advanced analysis of the earthquake recordings data, that determine the locations more accurately than a quick manual check, and reduce the location error margin in all horizontal and vertical directions.
For basic amateur usage, the first level manual verification is enough, but we have to keep the data uncertainties in mind at all times when doing the interpretation. Also of important notice is the earthquake monitoring network sensitivity, across Iceland in general, not just around Katla. The monitoring network is much denser today then it was 10 or 15 years ago, so we have to keep in mind that some earthquakes were not detected or located back in the days, which might in some cases across Iceland give (or amplify) an artificial appearance that the seismicity in some places is increasing compared to previous years. Compared to 5-10-15 years ago, the monitoring network sensitivity today is much greater and is thus also easier to accurately determine earthquake locations.
In this post, I am going to present some of my latest research on the Katla volcano. A lot was already said about its shallow activity and its shallow magma chamber. But I am going to go a bit deeper down, to a place where all the magic starts, where all the magma comes from. The place that is not mentioned or discussed very often. It is the very basement of the volcano, the magmatic roots, or as its also often called, “the deeper magma feeding system” or the “deep plumbing system”. I am just an amateur, and I have no formal education in this field, so this post will be written in a more “user-friendly” style, since I am not yet on a level high enough to throw around complex technical words. I am still deeply in the learning process, so I apologise in advance for any possible theoretical mistakes. Please feel free (and obliged) to point it out in the comments.
Build like a tree…
If we take a look at the seismic profile under Katla as a whole, we can separate it into several layers or “parts” as if it were a tree, which in a way it kinda is. And a very tall one. Now I do have to say at this point, that even tho the earthquakes are all manually verified (99% quality), there are still uncertainties regarding the depth, as mentioned before.
Based on the seismic data, Katla can be separated into 3 basic levels by depth:
– The shallow/upper system (2km and upwards) that includes the suggested shallow magma chamber, and all the hydrothermal systems, the glacier and the volcano itself above ground.
– The middle system (2-17km deep), where there are possibly many magma pathways, old dikes, sills, old magma pockets, and also a suggested secondary magma storage somewhere under the volcano.
– The deep plumbing system (17-32km deep), is where we have the main magma feeder root and where most of the magma intrusions from the mantle have occurred in the past few years, and probably also before in the past.
As the title of the post suggest, I am going to look closer at the roots of Katla, which is its deep plumbing system. Its seismic activity was low up until 2012. Since then, several individual intrusions have been recorded. An image of the past and present seismicity is on the top of the post.
Before looking at the roots of a volcano, we have to determine where they actually start. And that is a tricky part in itself, because we have to determine or find where the MOHO (Mohorovičić discontinuity) is. The MOHO as we know, is a boundary between the crust and the upper-most mantle. It can be up to several hundred meters thick. It is important to us, since it is from the MOHO or actually below it, that the magma comes into the plumbing system of a volcano.
If we could turn Iceland upside down, we would see the topography of the crust, which would look very much like mountains and valleys, through which potential magma from the plume can flow underground over a certain time period. On the image below that is seen. The deepest point is under the Vatnajökull glacier, and towards SE, closer to us, we see the deeper MOHO levels under the Mýrdalsjökull/Torfajökull area. This was a quickly generated graphic, just to show the basic underground topography under Iceland, based on the EUcrust07 model.
Katla is a volcano that is probably several hundred thousand years old, formed in the Pleistocene age. We can therefore rightfully assume that its feeder system (or also called “deep plumbing system”) is probably well defined and developed both at the MOHO and above/below it. So first we have to see where the MOHO actually is under Katla.
Many data-sets and studies exist that describe the MOHO under Iceland. Older studies (1996-2002) have all suggested the MOHO depth under Katla to be around 23-25km deep. But if we look at the seismic data, we can quickly see that earthquakes go all the way down to at least 29km. There is even one M2+ at around 31.5km from 2016. Earthquakes can/do in some cases generally occur also below the MOHO boundary, where the upper-most mantle above the asthenosphere is still solid enough to accumulate strain.
On the image above, we can see that the MOHO depth under Katla is not uniform by depth, but rises towards south. The colour bar on the right presents the MOHO layer depth. SW from Katla, it is around 30.5km and goes deeper down to around 33km and lower towards NE and beyond. In reality, the MOHO boundary under Katla is probably not so nicely flat/shaped as a sheet of paper in this graphic. But we can see that the bottom parts of the feeder root at Katla do tilt towards south, where the MOHO depth is a bit shallower than towards north where it goes down.
On the image above, we can see the seismic profile on the right (which is a view from east), that the deeper quakes abruptly end around 29-30km down. Why there is a lack of earthquakes between this “end” and the MOHO boundary can be explained in different ways.
First and most important, the MOHO model could be inaccurate.
Second, the very deep layers are on the border of the upper mantle, which is hotter than the crust and more ductile, so strain buildup is not so easily achieved as in the mid and upper crust.
Third, there could already be an open system from the upper-most mantle and MOHO up to 29km where the earthquakes start. This is not a young volcano and a lot of hot molten material was already transported from the mantle up through the deep layers up to the shallows, from where only a small percentage of the material actually erupted through the surface.
Fourth, the earthquake depth could be simply incorrect by a kilometre or two, since the deeper earthquakes are a bit harder to locate accurately by depth than the shallow ones.
And fifth, in reality the MOHO could be very non-linear, and could be more complex than a simple straight line, by having its own topography.
Now that we have seen where the doors to the basement lie, we can open it from below and move further up the system, up into the magmatic roots of the Katla volcano.
The magic number 3…
Just like the whole profile of Katla can be separated into a simplified 3-way system, we can also divide the main feeder root into 3 simple, yet well-defined layers, of the bottom, middle and upper part. And just like with Katla as a whole, we also need to take the earthquake depth uncertainty into account here. The separation of specific parts could be artificial, but even so, the feeder root is tall enough, that we can separate it into 3 areas anyway, because it is easier to track the progression of intrusion with height trough specific areas.
In this case, we can separate the root 3-ways simply because there are obvious areas of seismicity, separated by he earthquake swarm relocation or a seismic-free area. There are also possibilities of smaller sill formations, as the magma in the root rises and spreads also horizontally as it moves upward in the system. The whole system is not just one big open “tube”, but it is made out of many individual dikes and cracks and fractures. You could imagine it almost as a lightning bolt, which also has its “tentacles” spread out, like on the image bottom right.
Interesting to see, is how the main intrusions are confined to this root under the eastern caldera rim. other areas are almost aseismic, which is not so easy to explain. It could be that the material in that area is too ductile to accumulate strain, or the topography configuration of the bottom of the crust is such that fresh mantle magma is diverted straight into this main feeder root, which if its old enough, would be an open system at the bottom. It could also be that the crust in the surrounding area is too solid for the fresh magma, which tends to move through the weakest spots in the crust, which in this case would be the already established “dike swarm” (a non official term).
The general area of the root/conduit is around 3km across by a rough estimate. If we take it as having a cylinder shape, knowing its around 12km tall, we can calculate its volume at or around 85km3. This of course does not mean by far that there is so much magma in the root, but that is the approximate volume that the root occupies in its most centre area. Even if we assume that only 2-5% of the space occupied is actually fresh magma being transported, we still get a nice figure of around 2-4km3 of molten material present in the root at any time, being slowly transported into the higher systems over time, in the past and in the present. The actual amount/volume of fresh magma influx into the bottom system is not really known for Katla, and is also very hard to accurately predict from seismic data. Unless the transported magma starts accumulating at shallower depth, even GPS monitoring is of little help. The actual volume of transported material as well as the melt percentage can vary with depth/time, as can the age/freshness of the melt. The numbers above are just a very basic guesstimate, because it is hard if not impossible to actually accurately determine the real boundaries of the root or the individual parts of it.
The image above present seismic data from 2011-2017. The three blue dots that you can see between the middle and the upper part, are the 3 earthquakes associated with the intrusion event on April 13th, 2017. The first earthquake to occur was on the top of the middle section, and then the other two in the upper part followed. There was also a fourth one at the top of the bottom part at 24.5km, that followed a few minutes later. The whole event lasted around 8 minutes, and was the most well-defined deep intrusion since late 2015. The magma movement tremor was nicely recorded on the drumplots around Katla, and the tremor signal reached as far as Vatnajökull.
It is also hard to say how old exactly this feeder root is. We can however speculate whether it is an old pre-existing system, or a newly forming feeder dike. Current data indicates that this is probably a well established feeder dike, based on the speed of material transport (looking at individual dike progression speed), and the concentrations of earthquakes. With each intrusion, seismic energy release (earthquakes) can cover depths of around 5-10km by depth in a short period of time. If this were a newly forming feeder dike (or we could call it a “dike swarm)”, the earthquake progress upwards would be slower, and the energy release would be likely higher, as the rising magma would have to cut new pathways through the solid crust, which is not such a fast or quiet process. So it is safe to assume that this is a pre-existing, already well fractured feeder root, that has provided fresh hot basalt magma from the mantle in the past, to fuel eruptive episodes at Katla, either in basalt form, or as rhyolitic magma, altered by crustal storage/accumulation and magmatic differentiation. Given that it is situated under the eastern part of the caldera, it might also had its role in the run-up to the “Eldgjá fires” events.
I have analysed the seismic data, day by day, and I have put together a list of past deep intrusions under Katla. Some events lasted only minutes in one day and had only two or three earthquakes, while some could be tracked for a several days, with activity either moving with depth or with further activity in the same dike. Some earthquakes are probably not actually all related in a certain period, so this is more a list of deep activity events, rather than actual individual dike intrusions. The last example was the intrusion on April 13th, which was followed in 5 days by a new earthquake 2km higher (at ~17.4km) in the same area up along the root. It could be a separate event, but there is a high possibility that it is a late response to the intrusion, or just a continuation of it, like it happened in the past.
The list could use a revision/review, but as it is, it serves as a basic indicator of the activity over time. We do see a low amount of intrusion before 2011, which could also be due to the lower sensitivity of the monitoring network around Katla. There was some activity recorded nonetheless, which was more confined to the mid and upper system of the root. The activity in the bottom part has picked up substantially in the past few years.
It is obvious right away that the amount of deep activity has increased in the past few years. The intrusion that steps out right away is the one from 23.4-11.7.2015. Its progression and activity in the dike itself has lasted for two months and a half. It was nicely shaped, tilted with height and shaped like a true crack would look like, not completely linear, but shaped by the weakness in the crust.
When dealing with deeper activity, which is in most (if-not all) cases under Katla caused by magma movement, we have to look also at temporal progression with height, to see if the activity is stagnant or if there is any trend over time. In this particular case, a gradual upward progression would be expected and interesting to see. Which is why I made a plot of quake accumulation over the past years, 20 to be exact, which is a short period for a volcano like Katla, but it is the last 20% of its current repose time, which gives it some weight.
Obvious right away is the difference between the upper and bottom part of the root through the years.
From 1997 to 2011, we had mainly unorganised activity in the mid and upper parts of the root, and of course also some in the mid and mid-upper parts. Given that the period covers 15 years worth of data, the activity was mainly in background levels. There was probably more earthquakes, but back in those early years, the monitoring system was not sensitive enough to be able to locate all the deep activity, so there was probably some more, but the system has picked up higher magnitude quakes which account for the most energy release, which is low in this period. In 2008-2011, the monitoring system was getting much more sensitive, yet there was still no real organised deep activity detected. There was probably some magma movement in the bottom system at this point, but below the threshold to be able to trigger an eruption anytime soon if at-all with this pace.
In the 2012-2014 period, a substantial increase has occurred, especially in the bottom part. This was a first major sign that new magma is entering the system at a faster pace, since it was the bottom parts, closer to the MOHO, that were the most active, which is a clear signal of fresh magma entering the system from the upper mantle.
Lately in 2015-2017, the biggest difference has occurred in the mid and upper parts of the root, because we can see that the activity has picked up, and not only by numbers, but also by energy release. The dot sizes do have magnitude scaling applied, so we can see bigger magnitudes starting to occur in the mid and upper root system in the past years, and we see reduced numbers and magnitudes in the bottom part. This could imply that in the recent years the magma has been making more progress upward in the root towards the bowls of Katla. There is still some activity in the bottom parts, but given that the mid&upper parts of the root are picking up in activity over the years, it is an obvious sign that the intrusive melt is pushing upwards, with the bottom part likely being open or semi-open at this point. It does not mean that the magma is not coming into the bottom part anymore, but it could mean actually the opposite, that the bottom part of the root is now fractured or ductile from fresh magma, and unable to accumulate enough strain in the root itself.
The last deep intrusion on April 13th was also in the mid and upper parts, ranging from 24 to 19km, and a later quake was at 17.4km, which is still in the upper parts of the root. Also the activity just above the root has increased, to which we are going to get at a later point in future. We have to point out that the image in the center (2012-2014) contains full 3 years of data, and the image on the left (2015-2017) contains 2 years and 4 months of data. So we still have 8 months to go, and gather data, to have the same sample size as the previous period (2012-2014). We will see at the end of 2017 if it will look any different from now. This is also one of the most solid signs that Katla is indeed nearing an eruption in the coming months or years, since we can track the energy release going more focused from the bottom parts of the root, upwards into the mid and upper parts. Lately most intrusions are confined into these parts, with magma movement tremors now also being apparent on the drumplots.
What my personal research has shown, can be summarised into a few main points:
- The seismic activity deep under Katla, has been increasing fast since 2012, with many times more energy being released in the past 5 years than in the previous 15 years, and probably even further back. This deep activity is caused by fresh magma entering the system from the upper mantle.
- The deep activity under Katla is focused under the central eastern rim, and is horizontally and vertically confined to a specific area, which indicates rapid reactivation of a (likely) semi-permanent deep plumbing system or a magma feeding root.
- The magma feeding root is in average around 2.5-3km wide in diameter and at least 12km tall, with the upper parts ending around 17(+/-1)km deep, and the bottom seismicity ending around 29(+/- 2)km under the root. There is a strong possibility that the bottom end of the root is an open or a semi-open system down to the MOHO or below.
- There are clear signs of vertical magma transport in the root. The whole gradual reactivation process of the deep plumbing system since 2011, is probably one big slow ongoing event, composed out of individual smaller events. So while we were waiting to see large deep swarms ongoing for days or weeks and quickly coming to the surface, the deep plumbing system was already re-activated for a few years and is in an ongoing phase of transporting magma upwards into the upper systems. Like a slow but constant dike intrusion on a bigger scale.
- This reactivation of the deeper systems could be from the mantle plume pulse, which would be consistent with the data and the idea of a slower but steady constant fresh magma influx into the deep systems beginning around 2011/2012 and still ongoing.
- There is no apparent direct connection between the activity in the feeder root and the shallow activity at Katla. Also, the deep activity has no seasonal pattern to it, unlike the shallow activity.
- The reactivation of the feeder root is one of the most solid signals that the volcano is nearing its later stages of the pre-eruptive phase. Magma transport can be tracked all the way up to 16-17km which is on the border to the middle system, in which there is a likely complex of sills and dikes and magma pockets under Katla.
- There is a high possibility, even indicated by seismic data, that magma has already started to move into the middle system of Katla since 2013, after the stronger re-activation of the feeder root, but the GPS data is not consistent with this idea, which could be for various reasons, like the depth (10-16km) and the distance to GPS monitors, and the possibility of crustal wall assimilation, which could limit any potential inflation signal, before magma would get transported up into the shallow chamber.
- Re-heating and re-activating the entire volcanic system from the bottom up is a slow process, and we now have definite proof that the activation of the Katla volcanic system is already well underway for the past 5 years at least. Katla was not a cold system to begin with, and probably had several bigger intrusions since its last eruption.
- Before any eruptive sequence will start at Katla, I would expect seismic signs (earthquakes) of further magma transport in the system, in the depth range of 24-7km. As many have pointed out before me, the eruptive sequence at Katla is expected to be very noisy with seismicity, and preluded by uplift deformation, obvious on the GPS monitoring network.
- Personally I expect the next eruption to be somewhere in the eastern half of the caldera, possibly in the NE quadrant. The size of it will likely be from a strong VEI-4 up to a max of low VEI-5.
- In the company of Hekla and Grimsvotn, Katla is one of the most likely volcanoes to erupt in the near future, with an explosive eruption and with a potential hazard to aviation in the Euro-Atlantic sector in its initial days, when the ash cloud is tallest and contains the most fine ash grains.
- P.S.: During the 3D analysis of the region, no apparent connection was found between Eyjafjallajokull and Katla deeper and middle systems.
We have seen how the magmatic root of Katla looks like. But what lies just above it? Is there any indication of magma from the intrusions progressing upwards? Are there any changes under Katla compared to previous years, following the deep activity? I will try to answer these questions in a part 2, where I will ascend from the deep basement of the volcano, up into the first floor, into the very bowls of Katla, seen in the video below…
This is my first solo article for VolcanoCafe, and I would like to say thanks to the great “Dragons” of VolcanoCafe for giving me the opportunity to present my data and interpretation. This article only states my personal views and opinions.
Andrej Flis (Down Under – Andrew) for VolcanoCafe…