Chain of Dead Poets!

Amsterdam Island with visible craters.

The Amsterdam St Paul hotspot is one of the weaker hotspots around. It has created the St Paul and Amsterdam Islands, the now active Boomerang Seamount (last known eruption 1995), and an elongated chain of seamounts called the Chain of Dead Poets. These are remnants of the eruptive wake of the Amsterdam St Paul hotspot as the plates move on over it. The hotspot has had 2 episodes of increased activity after it became active. The first period lasted from 10 million years ago to six million years ago. The second period started 3 million years ago and lasts up until today. Amsterdam, St Paul and the Boomerang Seamount have all been produced during this second period of activity.

The hotspot is associated with the South East Indian Ridge and its rift system, and the chains volcanoes show evidence of changing in its chemical composition as the hotspot moved into the SEIR.

Amsterdam Island

The Island is the northernmost of the Antarctic sub-aerial volcanoes. It has had two eruptive centers down the line. Both with visible craters, the younger of the craters are far more visible on the image. Both of the craters are from periods of heightened activity, but later volcanism on the Island has primarily been of the flanking fissure type. Even though no eruption has been witnessed lava samples taken from the flanks of the younger crater shows that the volcano has indeed erupted during the last 100 years.

St Paul Island

The channel into St Paul natural harbour. One should keep slightly to the portside of the centerline of the channel when sailing in. The starboard side is much more shallow. By keeping slightly to portside of the middle you can get a 3 meter deep sailing ship into the natural harbour, well inside of it depth is not a problem, and you are quite safe regardless of weather. Stay away from the mammals on the beach, they are big and mean and are in no way to be compared to people in bikinis.

The island had a large eruption a few years before 1780 in which the predominant caldera formed. Even though the caldera is small for being a caldera it was probably formed by a Krakatoa style eruption starting with a for the volcanic system unusually large eruption with a subsequent magma chamber roof failure that let the ocean water down into the chamber. The ensuing steam explosion gutted the chamber.  In 1780 the vestigial remnants of the caldera wall facing the ocean crumbled and the ocean has during the following years carved out a fairly broad, but shallow canal that is open for smaller sailing ships due to its limited depth of around 3 to 5 meters.

Map of St Paul Island from Wikipedia. Note that the island is very small. The actual caldera is only slightly larger than 1 km across.

The Island is together with Isle du Kerguelen the best harbor in the southern ocean, and many trans-globe sailors make a port of call for repairs, or just general relaxation and landfall.

Boomerang Seamount

Not much is known about Boomerang that lies 18 kilometers north of Amsterdam Island. It rises 1 100 meters above the sea floor, but is still 650 meters below the ocean surface. During an expedition in 1996 they dredged up a lava sample and tested its Uranium/Thorium content. It showed that the lava had been erupted only 5 months prior to the visit.

The Seamount has a 2km caldera showing that the volcano has had at least one substantial eruption and probably have been a bit closer to the surface before.

CARL

The Dead Zone

Updated 13 Sep 2012, see end of article.

In logic, an assumption is a proposition that is taken for granted, as if it were true based upon presupposition without preponderance of the facts. (Wikipedia)

Back around May of this year, Carl asked me to do a series of simulations using KWare’s Heat3D, a program written by Ken Woheltz and the Reagents of the University of California under the sponsorship of the US Governement. It’s a cool little program that allows you to run heat simulations of magma intrusions into rock of varying characteristics. I had been prompted to write an article about one of the more perplexing areas in Iceland (well, to me it is). Not feeling that I was up to the task, I offered to do the supporting graphics if Carl could find someone to write the meat and potatoes of the article. I killed off a weekend working up the plots, but two of the catch points that we ran into were; “What temperature of the intruding magma should we use?” and “What exactly is the geothermal gradient of the region?”

With those two uncertainties, and the bedlam of real life, the post never made it to the forum. Things happen.

Before I go on, I must warn each and every reader here that I am not a seismologist, geologist, or bona-fide expert in the field. I read a lot, have been “studying” geology and physics in some shape form or fashion for about 37 years. I am just an amateur like many of you, so there is ample room for error.

With that out of the way… now we discuss

First, “The Dead Zone” is not an actual named place. It’s just a colloquialism specific to VolcanoCafe. It’s that region of Iceland between Katla/Torfajökull and Bárðarbunga/Grímsvötn. I refer to it as “The Dead Zone” due to the seeming low number of quakes. Historically, and pre-historically, the region is quite active with fissure eruptions. Irpsit and others can give you more definitive dates and names about the area, but I am limited to what I can cobble together from various sources. There are many other features here, but the main ones that I can find data on are Veidivötn, Vatnaoldur, Skaftar, Eldgja and Trollagigar. (spelling as listed in GVP and may be missing some of the diacritical marks) Veidivötn, Vatnaoldur, and Trollagigar are part of the Bárðarbunga system, Eldgja belongs to Katla, and Skaftar belongs to Grímsvötn. (As parts of the parent volcanoes fissure swarms). As you can see from the overview plot, there just are not very many quakes in this region. (Ignore the dot dashed blue line, that was part of the original plot set and is not used here)

Now, why is the Dead Zone dead? Because it is really… really hot. Much more than you would think. When an eruption is completed, magma sits and cools after the eruption is over with. This cooling rate depends on the thermal conductivity of the surrounding rock. For Basalt, the heat capacity is 840 J/kg K. (this is what I used in the simulations), Granite, for comparison is 790 J/kg K. This is in part due to its lower density. How it works… in order to raise the temperature of one kilogram of the material by one Kelvin (same as one degree C), you need 840 Joules of energy (for Basalt). Since we are talking about heat capacity, Water is 4185.5 J/kg K and Ice (at 0°C) is 2090 J/kg , so you can see how water or ice can drastically affect what is going on. This is one of those “gotchas” that can throw this whole scenario off. This area has a high water table and that can seriously affect how accurate the simulations are. Keep that in mind as I continue.

Anyway… when a dike intrudes into rock, whether it erupts or not, it starts loosing heat at a rate that can be calculated (provided you have the skill, or a program written by someone with the skill). Heat3D runs through the iterations of how heat migrates into the surrounding rock.
Here is how a single intrusion works out over a few years.

In my original set of graphics, I used a temperature of 1600°C magma due to the runniness of the flows and how far they traveled. My original guess was 1100°C based on a statement that I had seen in a paper, and much discussion occurred between Carl and myself about what would be the sane value to use.

“Time constraints on the origin of large volume basalts derived from O-isotope and trace element mineral zoning and U-series disequilibria in the Laki and Grímsvötn volcanic system” Binderman et al (2006) places the temp in the 1120–1140 °C range based on a “Mg in glass” geothermometer. (calculating diffusion and formation rates vs temp and pressure). Another reference (that I can’t locate at this moment) implies a temperature of 1200°C at 250MPa for one of the clast minerals. 250 MPa is in the 10 km depth range. Still uncertain of what temp to use, I went with the program default of 1250°C.

I used a 10 meter dike width based off of the average of three known dike sizes contained in “Geodetic GPS measurements in south Iceland: Strain accumulation and partitioning in a propagating ridge system” LaFemina et al (2005). This produces a really crappy 95% confidence range of 0.5 to 10.2 meters. (three samples is horrendous, but it’s all I had) Since the size of the plot grid has a direct play in how long the simulations take to run, I used 10 meters in order to get the simulations done in one evening.

Okay… now the actual run. As noted, this is not the original, and for brevity, I focused on only one system, Veidivötn. In case you didn’t know it, Veidivötn is probably the most lively fissure system in the region. It’s responsible for many of the Tungnaárhraun tephra layers. (THc. THd, THe…) GVP places an event there at the following dates: -6650, -4800, -4600, -4550, -4400, -4200, -1200, 150. For each eruption, I placed a 10 meter wide dike and ran the program out until the next intrusion date, which was then added and the process repeated. Another “gotcha” that you should be aware of, the eruptions did not necessarily occur in the same part of the fissure. This simulation assumes that they did. In effect, this skews the region towards being hotter than it might really be (and don’t forget the possible effect of the water that I mentioned previously)
So… here is the final product for what conditions may be like under the Veidivötn fissure. The temperature scale from the previous plot applies here.

Pretty gnarly eh? This is the crux of why I think that you won’t really see many small quakes in this region. Each one of those fissure lines has a heat structure similar to this. The crust is for the most part, plastic and yields to any stress that comes along… until it arrives too quickly for it to give. Then you have the larger quakes and potentially an opening of the fissure if the conditions are right… such as a nearby parent volcano being at or near erupting and having a ready supply of magma to flow down the rift and open it the rest of the way up. Structurally, there isn’t really much there to hold the two sides together. Plate shifts can do it (tectonic), or a parent volcano.

---

From “IAVCEI General Assembly 2008 Conference Field Excursions, Excursion 1: Historical Flood Lava Eruptions The 1783-84 Laki and 934-40 Eldgjá events” August 14-17 2008

“In 1783 the people of south Iceland had enjoyed a favourable spring and were looking forward to summer. However, their destiny was about to change. Weak earthquakes in the Skaftártunga district in mid-May were the first sign of what was to come. The intensity of these earthquakes increased steadily and on 1 June they were strong enough to be felt across the region from Mýrdalur and Öræfi. The earthquake activity escalated up until 8 June when a dark volcanic cloud spread over the district, blanketing the ground with ash (Figure 18a). The Great Laki eruption had begun.”

I’ve worked out the distances to Mýrdalur and Öræfi from the Laki site and applied an Mw to MMI estimate based on a few real world quakes from the USGS catalog in order to get a feel for how the power drops off over distance. Based on the MMI levels at which a quake becomes detectable by an unaided person, the quakes leading into the Laki event were in the Mag 4.5 to 5.0 range.

It’s a bit of a reach, but extending the formulas from “New Empirical Relationships among Magnitude, Rupture Length, Rupture Width, Rupture Area, and Surface Displacement” Wells and Coppersmith (1994) down to Mag 4.5, you get the following numbers.

Mag 4.5 – Surface rupture length 0.5 km, Subsurface rupture length – 1.3 km, Downdip rupture width – 1.7km.
Mag 5.0 – Surface rupture length 1.3 km, Subsurface rupture length – 2.7 km, Downdip rupture width – 2.9 km.

THESE ARE ESTIMATES

There is a bit of slop in the formulas, it is an attempt to get a working estimate of the physical manifestations that you would see from a quake. These particular formulas are only considered reliable for events down to Mag 5.2, but they do track well with no oddities in the curves. Below 5.2 the confidence in what the formula says drops off.

From that, it seems that the Mag 4.5 to 5.0 quakes are what is needed to open the system up. They have the right sort of features; the crust itself has likely healed very little from the previous events and should not take a lot of energy to re-open.

All this rumination and reading is one thing… but there is always something missing when you think and talk about these fissure eruptions. That’s the scale of the things. Since none of us were around, we just don’t know or have a frame of reference. All we have are eyewitness accounts. From some of those accounts, we know how long or how tall the fire curtain was, but that’s it. Just numbers in a book. Here, I have scaled an image of a generic fissure eruption and placed a few well known silhouettes in front of it so that you can see just how big these things are.

Enjoy.

GEOLURKING

GL Edit: The silhouetted buildings are;
Empire State Building – 443.2 m, Taipei 101 – 449.2 m, Burj Khalifa – 829.84 m, Sears Tower – 527 m, Petronas Towers – 451.9 m

“GVP” = Smithsonian Institution – Global Volcanism Program

UPDATE:

Irpsit says:
September 12, 2012 at 18:26

From what I know Laki eruption could be observed from almost anywhere in Iceland, in distance. The reports even speak that you could see the fountains from far away, but probably not everywhere in Iceland, as 1km high is not enough for that.

This put me on a search for two of the images that I made for the original article. I was able to pull them from Google archive of my mail.

They are not as stunning as the scaled image, but they are worth pondering. The ruddy maroon rectangle represents the Skaftar (Laki) fire curtain anchored to the surface, as seen from a couple of locations.

El Hierro – Day 3!

Photograph by photosaereasdecanarias.com Tanganasoga volcano, one of the possible places where the new eruption will occur.

Today we have seen a change in the behaviour at Pevolca, now they have changed the level of alarm from green all over the island, to yellow for Julan and La Dehesa. They are also asking everyone on the island to familiarise themselves with the plans for evacuation and self protection.

Image courtesy of IGN for Volcano Café. The image is showing the Seismograph station of Julan. It was during yesterday giving the highest values of all the stations on El Hierro. The center of the probable source yesterday was in the area from Tanganasoga to Humilladores.

The center for the current magmatic earthquake swarm has moved slowly since yesterday to the west and south. And the magmatic component of the waveforms have increased in clarity. As you can see in the image above (provided directly by courtesy from IGN to Volcano Café) the center of attention yesterday was at Julan.

There has also been rapid inflation during the last two days with clear movement patterns associated with magmatic movement, and new magma arrival. The numbers are high for such a short time period, between 2 and 3 centimeters.

Commentator Vishy pointed me to a paper by Stroncik regarding chrystalization in magmatic chambers, and in this case the depth necessary is somewhere between 16 and 23 kilometers down. This means that the earthquakes we have seen in this depth region is where the magma chambers are.

Image by IGN. As you can see the band of red and blue is moving to the southwest. The area is showing part of the magma chamber system under El Hierro.

As you can see in the image this increases the risk for an eruption occuring in the southern fissure zone ranging from La Restinga up towards Tanganasoga. If a new feeder channel is opening up we will first see migratory earthquakes in the region between 16 and 8 kilometres, after that we will most likely see an onset of eruption within a few hours. It is therefore very important to study carefully the plots to see where the conduit is forming.

The likelihood of a new eruption starting is now very high. I would like to tell any readers of this blog who lives on El Hierro to check often for new data, especially the IGN Earthquake positioning map. If you live in the area ontop of where the earthquakes are forming in the 16 – 8 kilometer depth level, you would do well by moving yourself away to safer grounds, best and safest part is to the northeast on the island (very low risk for eruption there). You are of course welcome to check in here also for new information in the comments, if anything happens it is most likely to find the information you need there.

http://www.01.ign.es/ign/resources/volcanologia/jpg/Eventos_HIERRO_2D.jpg

Image by IGN. A nice and very clear image of a Longperiod Earthquake (LP event). Thanks to commenter Judith who pointed it out for me.

CARL

What’s going on at Katla? Part III

Image from Wikimedia. Aerial picture of Katla.

Trying to make sense of complex phenomenae

In the first two instalments, we had a look at Katla as she appears through media and what she has done historically. It is now time to have a look at what’s going on and try to paint a coherent picture of what she actually is, is up to and able to do, but first let us recapitulate what we found previously:

• There is a general interest in Katla because she is and has been regarded as a very dangerous volcano by generations of Icelanders.
• The presentation of Katla in media is skewered by vested interests ranging from scientists who hope to increase their professional and/or public standing, people trying to cash in on the interest generated such as journalists and bloggers, and finally, there are people trying to increase their standing within the subculture of doomsaying and alarmism.
• Katla is a massive but relatively young volcano, located on the MAR, and formed when Iceland was covered by glaciers.
• The records include two large fissure eruptions on the NE flank of Katla; the prehistoric 5 km3 Hólmsá Fires of 5550 BC and ~22 km3 Eldgjá eruption in 934 AD. In historic times, the 1100 years or so that Iceland has been settled, there have been 27 listed eruptions (28 if the inferred minor subglacial 2011 eruption is included), 23 of which have been explosive.
• Of the 23 explosive eruptions, three have been assigned VEI 3, thirteen VEI 4 and four VEI 5.
• The four VEI 5 eruptions are remarkably alike in size at 1.2 – 1.5 km3, which is at the upper end of what Katla probably is able to do but at the very lower end of VEI 5 eruptions.
• Tephrochronology (in some cases complemented by radiocarbon dating) has identified a further 103 eruptions going back ~8,500 years, and in the few cases where a VEI has been assigned, none have been greater than a VEI 4.
• Katla does not possess a caldera-sized magma chamber.
• In order to account for the great number of explosive eruptions which involve more evolved magmas, Katla could have more than a single magma chamber.
• The available evidence suggests that in order to break through the up to 700 meters thick Mýrdalsjökull glacier, an eruption must be at least a substantial VEI 3.
• Direct and (primarily) indirect evidence suggests that smaller eruptions, mainly basaltic VEI 0 – 2 eruptions are severely underrepresented in her eruptive record and ought to exceed the number of observed eruptions.

Fig 1. Mýrdalsjökull showing the main glacier outlets, directions of jökulhlaups and areas affected. E –
Entajökull, S – Sólheimajökull, K – Kötlujökull, M – Markarfljot, Ss – Sólheimasandur, MS – Mýrdalssandur.
Eyjafjallajökull is to the left and the smaller glacier above is Tindfjallajökull (adapted from Google Maps).

The greatest danger from Katla comes from the very quick and extensive melting of the glacier caused by large eruptions which results in destructive jökulhlaups. There are three major outlets from the glacier: Entujökull to the NW that empties into the Markarfljot river and valley north of Eyjafjallajökull, Sólheimajökull to the SSW that empties onto the Sólheimasandur and finally, Kötlujökull to the SE that empties in a great arc east through south onto the Mýrdalssandur. What ought to be prime farmland and in fact once was settled, is nowadays an unsettled wasteland because of the devastating jökulhlaups unleashed by Katla. This is the true reason why Katla is considered to be such a dangerous volcano.

The fact that one often comes across the reference that in the days before the Hringvegur (ring road), “people were afraid to traverse the Sólheima- and Mýrdalssandur because of the frequent jökulhlaups” is another indication that smaller and unrecorded eruptions that cause only minor hlaups are far more frequent than the 40 – 80 years often given as the interval between main, and thus visible, eruptions.

Fig. 2. The foundations of the old bridge across the Múlakvísl river destroyed by the July 9th 2011 jökulhlaup
are visible to the left. The new bridge was laid down a week later. (photo John A Stevenson, GVP website)

Apart from the postulated connection between the Eyjafjallajökull and Katla volcanoes, one question that always crops up is the Goðabunga cryptodome. Many volcanologists maintain that it is a part of the volcanic system of the Katla central volcano. Others, notably Sturkell and his co-workers, claim it is part of the Eyjafjallajökull volcanic system. In order to shed some light on this issue, I asked our own GeoLurking if he could make a plot of all the earthquakes from 1994 up to and including the 2010 Eyjafjallajökull eruption. The results are quite clear and do throw up a surprise:

Fig 3. E-W cross section, view from south, through Eyjafjallajökull, Goðabunga and Katla. Plot by and
courtesy of GeoLurking. The “lines” formed at approximately 5, 3 and 1.1 km at Goðabunga and Katla are most
likely artefacts caused by quakes being assigned a poorly defined depth. The latter, 1.1 km, is the default depth
assigned by the automatic system in case it cannot compute a depth within the predetermined level of certainty and unless they are manually checked, which is not the case of every quake, automatic depth remains uncorrected, hence these artefacts.

From this cross section, it is quite clear that there is no connection between the Eyjafjallajökull volcanic system and Katla. Eyjafjallajökull has its own, well-defined feeder system from the Moho (first molten layer beneath the Earth’s solid crust) as does Katla, thus they are wholly independent of one another. As can also be seen, albeit not as clearly, Goðabunga too seems to be independent of either Eyjafjallajökull and Katla, the ramifications of which will be the subject of a later post by Carl. Sufficient to say that when we contemplate what Katla herself may be up to, we must differentiate between activity at Goðabunga and activity at Katla. Once we do, we see that while Goðabunga is more or less continuously active, Katla operates in bursts and seems to be most active during summer and autumn when the ice cap is at its, relatively speaking of an up to 700 m thick glacier, thinnest.

Fig 4. Activity post-Eyjafjallajökull. Activity at Eyjafjallajökull is minor and has to do with the system settling down after the end of the eruptive phase. Note that at a depth of 0 to 5 km or so, there seem to be three separate areas of activity at Katla. (Plot by and courtesy of GeoLurking.)

After the Eyjafjallajökull eruption, Katla seems to have entered an active phase with a suspected subglacial eruption on July 9th 2011 and several pits or craters forming on top of the glacier. This activity seems to be localised to three main areas within the caldera:

Fig. 5. Earthquake activity at Katla July 9th 2011, the day of the jökulhlaup and suspected subglacial eruption. Both the 1823 and 1918 eruptions occurred close to but just east of this area. The 1823 eruption occurred close to the easternmost red spot while the 1918 eruption was roughly at the rightmost dark blue spot below it. (IMO)

Fig 6. Earthquake activity at Katla July 17th 2011. (IMO)

Fig. 7. Earthquake activity at Katla July 21st 2011. The 1755 eruption was situated in the same area as the three overlapping orange spots. (IMO)

As can be seen, there are at least three distinct areas of activity inside the caldera with the one associated with the inferred July 9th eruption well to the south. The pits formed in the glacier also align with these three areas, albeit the pits to the northeast seem more drawn out along the caldera wall and not over the center of activity. These three areas seem to tie in with the three areas of activity noted in fig 4 as do the locations of three of Katla’s major eruptions. Thus there is not a single vent, but at least three at surface distances of approximately 5 to 8 km from each other. It is equally likely to judge from Fig 3. and Fig 4. in conjunction, that at great depth, they do have a common source.

I will now present you with my personal view of Katla, but do not be afraid to disagree or draw your own conclusions (within reason please, no Katlatubos here):

Katla is a young volcano and far more active than has previously been thought. Unlike the similarly aged but much less active Eyjafjallajökull, Katla has had more time to develop her system of sills to the point where they are fewer in number than they originally were but have a substantially larger magma-carrying capacity and approach or may have reached the point where they can be considered magma chambers proper. Since cooking evolved magmas takes a long time, usually millennia in the case of cubic kilometre-sized silica-rich magmas and at the very least many centuries for intermediate magmas, it is highly likely that Katla possesses several pockets of magma capable of eruptions ranging from high VEI 3s to small VEI 5s. Not only do the times between such eruptions argue this, their wide spread of location within the caldera does so too.

The most common type of eruption at Katla is the small, subglacial eruption of a few tens of millions of cubic meters of basaltic magmas. These eruptions are not energetic enough to break through the very thick Mýrdalsjökull glacier and the only proofs of their existence are intense earthquake swarms followed by minor jökulhlaups and later observations of deep pits or craters, sometimes water-filled, in the glacier ice. My guesstimate is that there may be many such small eruptions over any given ten-year period, and possibly in the case of a period of high activity, there may even be more than one in a single year. By back-tracking and investigating old accounts over the past few centuries of jökulhlaups in the area not associated with visible eruptions, it ought to be possible to identify many of these minor eruptions.

While a larger “proper” eruption of Katla in the VEI 3 – 5 range cannot be ruled out, I find one unlikely at present as the current activity mostly is in areas already depleted of evolved magmas by geologically speaking very recent major eruptions. Also there is little sign of the uplift required on GPS. If one were to occur, the odds for one towards the upper end of what Katla is able of ought to be better in the Eastern to Northern parts of the caldera.

Finally, what we do see when we look at SIL-stations such as Austmannsbunga, located on the NE caldera rim (not a coincidence, see above), is hydrothermal activity following a period of possibly still ongoing magmatic intrusion and not signs of an imminent, large eruption.

Fig 8. Hydrothermal activity at Katla as shown on the Austmannsbunga SIL (IMO)

I’m sorry to be such a boring old fart, but if this is unsatisfactory, start looking for intense earthquake activity at some 25 – 10 km depth, showing on the IMO map for Mýrdalsjökull as being in the Eastern to Northern part of the caldera, activity that shows a clear upwards trend and spreads when it reaches depths approaching 5 km!

HENRIK

What’s going on at Katla? Part 2

Part 2, A view of Katla

Fig. 1. Katla from Háfell looking NNW (RUV webcam capture)

So what really is going on at Katla? Well, we’re not really there yet. In this instalment, I will summarise what I have learnt from reading various scientific or otherwise papers and articles and my current understanding of it. At certain points I will supplement this with what I believe to be or could be the explanation, but when I do, I will say so. Again, I emphasise that I am not an expert in any way.

Katla is a relatively young volcano which like so many Icelandic volcanoes formed when Iceland was covered by ice. Hence it is a tuya, steep-sided with a broad, flat top. Like other large Icelandic volcanoes, it has a very large summit crater described as a caldera, but one that did not come about as a result of the collapse of the volcanic edifice into an emptied and very large magma chamber as happened at Mount Mazama a.k.a. Crater Lake in Oregon, at Krakatoa or at Long Valley.

Fig. 2. Herðubreið, a subglacially formed tuya with steep sides and a flat top. Post-glaciation, erosion has
made the sides less steep and a small post-glacial cone makes the top appear less flat than it once was. The
similarity to Katla, once you allow for the vast differences in size, is obvious. (extremeiceland.is)

One of the keys to understand what goes on at Katla is to have an idea of what lies beneath the up to 700 meters thick glacier that covers her crater/caldera. In schematic representations of Katla, a magma chamber at the very shallow depth of three to five kilometres is often displayed. From reading descriptions of other volcanoes that have suffered caldera collapse or looking up a general definition of ”caldera”, it is easy to assume that Katla too must have a magma chamber that spans the entire width of the “caldera” and which, “once-upon-a- time” collapsed to for the present-day caldera. Nothing could be further from the truth, but alas, there is no direct information available that accurately describes what Katla’s magmatic system, the true volcano, looks like. We have to fill this gap ourselves.

The first thing to do is to look at what she has done in the past. If we look up her “Eruptive History” on the Smithsonian Global Volcanism Program website, we find that Katla is listed as having had 27 eruptions during the period Iceland has been settled by humans, some eleven centuries and counting. Of these, only the larger eruptions seem to have been registered prior to the middle of the 20th Century. Thus the 27 eruptions are divided as follows: Two VEI 0 (1955 and 1999), three VEI 3, fourteen VEI 4 (including the AD 934 “Eldgjá fissure eruption”) and four VEI 5 with a further four not assigned a VEI number. Of the four unassigned eruptions, one is listed as “subglacial, lava flows” and three “subglacial, explosive”. Please take note of the dearth of smaller eruptions, VEI 0 – 2, as this is important and something we’ll return to later.

From this information, it is clear that Katla cannot have a single, caldera-sized magma chamber because such a chamber would contain several tens to even hundreds of cubic kilometers of magma, which in turn would have led to far larger eruptions. None have occurred. Since VEI 5 is assigned to eruptions that eject between 1 and 9 cubic kilometres of Dense Rock Equivalent (DRE) explosively, and Katla’s VEI 5 eruptions are remarkably consistent at between 1.2 and 1.5 cubic kilometres, anything much larger than some 3 – 4 cu km is rather out of the question. A caveat – given the area covered by the crater/caldera, there could be more than one such chamber responsible for her eruptions, in which case it would be fair to ask the question if Katla really is a single volcano or if not a description of her being several volcanoes rolled into one would be more accurate.

If we look at her eruptive history prior to Iceland being settled, deduced by tephrochronology – ash layers deposited being identified by their physical properties, such as chemical composition and grain size, as belonging to Katla and from the size, distribution and time derived for each individual layer of tephra, an eruption responsible for it is inferred – we find that there have been a multitude of eruptions, but only a few of which have been assigned a VEI number. Interestingly in every such case a VEI 3 or 4 has been deduced. Anything much larger must have left such extensive deposits that such a huge eruption cannot have escaped detection, hence we can conclude that no explosive eruptions larger than a small VEI 5 have ever occurred at Katla.

There have been two exceptions to the rule that Katla’s eruptions normally are in the VEI 4 range volume-wise. Both originate on her NE flank, outside the crater/caldera. Around 5550 BC, Katla was the source of the 5 cubic kilometres “Hólmsá Fires eruption” lava flow. In 934 AD, the four times larger “Eldgjá eruption” spewed forth some 18 cu km of lava and five cu km of tephra, or ash. Even if the total volume erupted in 934 AD, about 22 cu km DRE, is on the order of 50 times greater (25 to 200 times), a lowly “VEI 4?” has been assigned.

As the underlying causes and processes that drive “regional fissure eruptions” are vastly different and as they happen very rarely, seemingly with a time interval measured in several millennia in the same-ish location, fissure or rift eruptions should be considered separately – even if the visual appearance of the Katla crater/caldera suggests that a fissure eruption has at some point in the distant past intersected it. They are mentioned here because an article such as this cannot fail to do so, nor can it fail to give a reason why they are not included in the discussion.

Earlier I mentioned the apparent absence of small eruptions from her eruptive record with only two “possible subglacial eruptions” in 1955 and 1999 listed, to which can now be added the equally suspected or “possible” July 2011 subglacial eruption. As I write this, it seems that there may have been yet another, very minor hlaup. That such eruptions were not noted in earlier days is not surprising as the very small hlaups they resulted in were local nuisances rather than regional catastrophes of a major Katla jökulhlaup and would not have been seen as important enough to be recorded, even had they been observed. But how frequent could this type of small eruption be?

Fig 3. Seljansfoss Waterfall during the 2010 Eyjafjallajökull eruption (Binaural Waves Blogspot). Notice
evidence of several minor eruptions on the mountainside above the waterfall.

We know from the 2010 Eyjafjallajökull eruption that it was preceded by two fissure eruptions at Fimmvörduhals that intersected each other. If we look at the topography and geography of Eyjafjallajökull, we can see many areas of monogenetic cones. This indicates that eruptions of the Fimmvörduhals type greatly outnumber eruptions at the main vent. At Askja, a similarly sized volcano albeit glacier-free and with a slightly smaller summit crater/ caldera, there have been six small eruptions since the great eruption of 1875 and many prior.

Of the 24 eruptions (not counting the AD 934 Eldgjá fissure eruption) listed before it was realised that there were smaller eruptions that would only show as minor jökulhlaups, 20 are listed as VEI 3 or higher and three of the four not assigned a VEI number are listed as (subglacial and) explosive. At least 17 of the 23 explosive eruptions have been assigned a VEI of 4 or 5. The eruptive record of Katla thus indicates that in order to break through the up to 700 meters thick Mýrdalsjökull glacier, an eruption would need to be at least as powerful as to merit a designation of VEI 3. Thus – the reason for the dearth of smaller eruptions observed is that they are not energetic enough to break through thick glaciers such as Vatnajökull or Mýrdalsjökull to be visually obvious and the minor hlaups resulting have been much too insignificant to have been considered as a result of an eruption that never was seen.

Fig. 4. Pits formed by melting from below in the Katla glacier, summer 2011. The glacier was still covered
with tephra from the Eyjafjallajökull eruption which made such features stand out unusually well.
(ModernSurvivalBlog, picture may originate with Icelandreview)

With the advent of aircraft, it was noted that there were pits in the glacier as if it had melted from below and the collapsed to form an ice crater. These pits are relatively numerous and vary in size. They have been explained as due to either strong hydrothermal activity or, in the case of the larger ones, as the result minor subglacial eruptions.

The obvious conclusion is that in the case of Katla, small eruptions of the Fimmvörduhals type far outnumber the bigger, recorded eruptions. This is vital for understanding how a volcano such as Katla is built and works.

Let us for a moment return to what I like to call “Katla’s defrosted twin”, Askja. Here we can see, side by side, the effects of the two types of eruption. In 1875 she had the big VEI 5 eruption, about four times as great as Katla’s historic VEI 5s, that would eventually form lake Öskjuvátn. Here we have a magma chamber where magma collected over time, partially re-melting and absorbing the chamber walls which together with fractionating led to the body of magma collected being far more silicic than the basalt injected into the chamber, which provided the heat or energy for the process. This went on for centuries, quite likely millennia as GVP lists the preceding very large eruption at Askja as having occurred about 11,000 years ago, until a final basaltic intrusion was energetic enough to unbalance the magma chamber and the big eruption of 1875 followed. Please note that both before and after, there have been many smaller, basaltic eruptions that have evidently bypassed the main magma chamber on their way to the surface, one of which caused the miniscule crater Vítí located immediately north of Lake Öskjuvátn.

Fig. 5. “Katla’s defrosted twin”, Askja. Aerial photograph inside and above the Askja caldera with Lake
Öskjuvatn and the miniscule crater Viti barely discernible on the near left-hand side of the lake. (uwmyvatn
blogspot)

This too is what I believe must have been happening and is going on at Katla. Sturkell and his co-authors in their 2009 paper “Katla And Eyjafjallajökull Volcanoes” note that the products of Katla’s eruptions are bimodal, comprising alkali basalt and mildly alkalic rhyolites “with intermediates very subordinate”. One, or possibly more magma chambers where magma collects, fractionates and grows more silicic, a process that takes hundreds if not thousands of years which is why more than one magma chamber seems to be required in order to account for the relatively frequent eruptions of Katla, until there eventually is an eruption of “mildly alcalic rhyolites”, accompanied by tens to hundreds of smaller, alkali-basaltic eruptions which due to their location under the ice in a watery environment, gouge out small craters and fill in the bigger ones with mostly small, broken fragments of lava, piles of pillow lava or even small lava flows or easily eroded cones. When a big eruption occurs, the glacier first closes the wound, then the crater gets back-filled with loose rubble which gets pasted over with more solid lava flows from later eruptions.

This process has been going on for as long as Katla has existed. Not only has this constant remodelling inside the crater/caldera left a kilometres-deep zone of clastic, i.e. broken or fragmented, rock mixed with water, it also in my opinion explains how the caldera was formed in the first place. This layer extends down to not much above the roof/-s of the magma chamber/-s. As freshly injected basalt from the mantle makes its way up, it will eventually encounter this water-rich zone and result in intense activity, hydrothermal at first, and if the intrusion continues, hydromagmatic. It is primarily this activity we see when we look at the tremor charts of the SIL-stations surrounding Katla, in particular the one located at Austmannsbunga, on the north-eastern crater/caldera wall.

In the next instalment, it is time to take a look at Katla’s neighbours Eyajafjallajökull and the Gódabunga “cryptodome” and try and separate their activity from that of Katla so that we can finally figure out what she may have been up to over the last few years and how likely an eruption in the near future could be.

HENRIK