Bárðarbunga – everything you need to know (well, almost). 21 August 2014   5 comments

Yesterday a colleague decided to hold a ‘mock’ interview with me during our lunch break, which she recorded and I’ve just written up and tidied up. You may find it informative.

Q. So Dave, stop the 50:50 stuff when asked ‘will it erupt’. What do you really think?

A. It’s still 50:50! Whether it will erupt or not depends on a number of factors, some of which cannot be monitored. So that people can better understand why predictions are so difficult let me list some:

The magma is sitting at depth in a vertical fissure and slowly moving NE. It’s a dyke intrusion.

A key question is whether new magma is joining the magma in the dyke. If not (or it’s just a small amount), then there is unlikely to be an eruption. It will stall and cool.

However should a fracture suddenly appear above the dyke, then the magma is going to move upwards, and then it’s more likely to erupt.

Because as it moves up, it will reach a level where any dissolved gases (mostly water) will stop being dissolved, expand dramatically and accelerate upwards, and ‘push’ the magma to the surface. This is actually how eruptions are powered – bubbles.

Another scenario is if magma keeps being pumped into the dyke. The dyke has a number of choices: use the extra energy to keep moving NE; expand by moving to the SW, or grow up and/or down.

Get the picture?

Q. Thanks Dave, and stop calling me Bubbles. Right, we all love an apocalyptic story, so what’s the worst case scenario?

A. Ah, well, there’s more than one with this particular volcano – sorry. But these are nowhere on the horizon at the moment. Here are three.

  1. This presently benign little dyke intrusion is the forerunner to the uprise of large packets of melt from below (from the mantle) and it suddenly turns into a Laki-type flood basalt eruption. There’s still controversy over how these massive eruptions are fed in Iceland, but they always occur in fissures, and they have to involve the mantle because we have no definitive evidence that 10s of cubic kilometres of melt are stored under each central volcano just waiting to erupt. A little puzzle to solve is why these flood basalts (if they are fed directly from the mantle) have ‘shallow’ pressure signatures, but this might just mean they spend enough time at shallow dept in transit to ‘equilibrate’ to lower pressures.
  2. This event triggers activity within the heart of Bárðarbunga, beneath the summit, where there’s almost certainly some melt and or mush (melt+crystals) stored. This could be all basalt, or there could be some more ‘sticky’ magma around, such as rhyolite. Evidence from ash layers in Iceland indicates that explosive basalt eruptions from Bárðarbunga do happen, and that they are powerful. The good news is – and myself and John Stevenson have said this many times – is that we have less to worry about if this happens because we’ve already had one – Grímsvötn 2011. So we know that fewer flights will be cancelled simply because the old “ash in the sky you don’t fly” rules no longer exist. Everyone is much better prepared for a big and powerful explosive eruption. I’ve seen a few geologists say things like “Icelandic magmas do not contain enough gas to drive powerful explosive eruptions”. This is utter rubbish, incorrect, and misleading. These are invariably geologists who lack a true understanding of Icelandic volcanism because they have done little or no research there.
  3. Probably the worst-case scenario for Iceland is that this leads to a massive volcano-tectonic event in the fissure system to the SW of Bárðarbunga, as this is where a number of large flood basalt eruptions have occurred. The hydroelectric power plants on the rivers near to this fissure system would be in trouble, and we know that in the past large ash piles have dammed the rivers. The abundant water in this area results in spectacular (but fairly local) explosions and a high production of fragments as the abundant river water cools the erupting magma.

Q.Final question. You mentioned over coffee that you’d been very active on Twitter trying to get what you called the ‘right information’ out there. But isn’t there a danger that others will pinch your work and re-cast it as their own?

A. That comes with Twitter territory. I’d much rather try and provide an informed and scientifically-based set of views and ideas that can be pillaged and re-used (usually without credit) than leave it to those who don’t understand Icelandic volcano-tectonics to mislead (not always deliberately I hasten to add). I appreciate it when folks give me credit, but I don’t expect it. If you are being paid from public money to do your science, then put your knowledge to good use for the benefit of the public. Getting credit for it is a bonus, not a right.

Q. OK – late for the next meeting Dave. Maybe continue with a pint or two later?

A. Only if it’s a real ale acceptable to my palate.

Posted August 21, 2014 by davemcgarvie in Uncategorized, Volcanism

Askja Volcano – 26 July visit to the 22/23 July 2014 Landslide   1 comment

On the night of 23/24 July 2014 (around midnight) there was a large landslide in the SE corner of the steep inner wall of the 1875 AD caldera at the Askja volcano in central Iceland. This event is simply the latest (albeit large and spectacular) of many that have formed the current water-filled caldera of Öskjuvatn (Askja lake). It is part of the ongoing process of the formation of this youngest caldera at Askja, which is after all only 139 years old and which after its initiation in 1875 took several decades (until c.1932) to get close to the shape we see today.

This blog post contains images from before and after the landslide. I was fortunate to be doing fieldwork nearby (collecting samples from basalt subglacial mountains) when I heard that access to the ‘safe’ area above the lake had just been granted. So we went there on the first day that the area had been opened since the landslide. It was a bit special.  A. Takn in July 2011 - the site of the July 2014 landslide.

A. Taken in July 2011 – the site of the July 2014 landslide.

B. Taken in July 2011. Yellow line shows the major fracture system that was exploted during the July 2014 landslide.

B. Taken in July 2011. Yellow line shows the major fracture system that was exploited during the July 2014 landslide.

C. Taken in July 2011. Purple shaded area shows roughly the part of the inner wall that collapsed during the July 2014 landslide.

C. Taken in July 2011. Purple shaded area shows roughly the part of the inner wall that collapsed during the July 2014 landslide.


D. Taken on 26 July 2014, 3 days after the landslide.

Estimates of the volume of the landslide range from 24-60 million cubic metres, and no doubt this will become refined as Icelandic scientists either gain access to the area or use digital elevation models to obtain more precise measurements.


The main hazard from the landslide was not the slide itself, as this occurred in a location well away from tourist trails. This location is visited only rarely by geologists utilising the superb exposures revealed by the caldera collapse to gain deeper insight into Askja’s past geological evolution. See Graettinger et al., 2013.

Nope, the main hazard from the landslide was the wave triggered by the sudden entry into the lake of a large mass of debris. Various people have called it a tsunami, a displacement wave, and a seiche. Tsunami will do, and estimates place it as 60-75 m high when it reached the opposite caldera wall there the vast majority of tourists gather to gaze over the lake and into the small crater of Víti (Hell) filled with turquoise coloured, warm, and sulphurous water. Fortunately nobody was in Víti at the time or they’d have had a shock (and an unwelcome cold shower) as the top of the tsunami wave spilled into Víti.


The small water-filled crater of Viti, which lies just north of the rim of the 1875 AD (youngest) caldera at Askja. It was one of the vents of the 1875 eruption - the rest are buried beneath the lake water.

Figure E. The small water-filled crater of Viti, which lies just north of the rim of the 1875 AD (youngest) caldera at Askja. It was one of the vents of the 1875 eruption – the rest are buried beneath the lake water. ‘Spillover’ marks the low point where water from the tsunami wave poured into Viti.



The image below shows the raft of rhyolitic pumice and ice that remains after the landslide, with the source of the pumice being loose and unconsolidated deposits from the 1875 eruption. It will be interesting to see how long this raft persists, as the strong winds of Autumn and Winter will deposit much of the material on the eastern shore.

Raft of rhyolitic pumice and ice occupying the northeast corner of Askja lake. Debris-covered areas clearly indicate inundation bythe tsunami wave. Access to these areas to check extent of inundation was not possibe as the area is closed.

Figure F. Raft of rhyolitic pumice and ice occupying the northeast corner of Askja lake. Debris-covered areas clearly indicate inundation by the tsunami wave. Access to these areas to check extent of inundation was not possible as the area is closed.


Images from the landslide source – 2010 and 2011

August 2010, at the eastern end of the headwall of the July 2014 landslide. Outcrops show downward movement relative to the ridge crest, and multiple parallel troughs indicating fault development.

Figure G. August 2010, at the eastern end of the headwall of the July 2014 landslide. looking to the west. Outcrops show downward movement relative to the ridge crest, and multiple parallel troughs indicating fault development. Some block rotation resulting in dip to the south (left) was apparent on closer inspection.

In 2010 and 2011 I was co-supervising a PhD student who was mapping the older rocks that lie on the east and south of the young 1875 AD caldera right down to the lower outer flanks of the volcano. I also visited the southeast corner with an Earthwatch group in 1985 and made the surprise discovery that there was an old rhyolite dome here, which I confirmed with a chemical analysis. It was apparent that the area around the top of the rhyolite dome and to the west was unstable and that a fault system had been active given that parts of the rhyolite dome had moved downslope and been rotated to dip 5-15 degrees to the south.

From the eastern end of the 2014 landslide headwall, looking across the lake to the Viti crater.

Figure H. From the eastern end of the 2014 landslide headwall, looking across the lake to the Viti crater.

To be honest, on the past occasions I was above the headwall of where the July 2014 landslide occurred (i.e. in 1985, 1987, 2010, and 2011) I was aware of the potential for a landslide in this area, but from the evidence I could see of other landslides (especially the older one immediately to the east) it looked like any future landslide may be a gentle slump rather than a headlong dash into Askja lake.

Older landslide immediately to east of July 2014 landslide. This older one contains large intact blocks of rotated rhyolite lava from the dome above. Look carefully and you can see these be seen on images A-D above.

Figure I. Older landslide immediately to east of July 2014 landslide. This older one contains large intact blocks of rotated rhyolite lava from the dome above. Look carefully and you can see these be seen on images A-D above. In the foreground is one of the basalt vents from the 1920s, when a number of basalt (and mixed-magma) eruptions occurred around the 1875 caldera margins. This vent erupted a number of silicic lithics (non-juvenile clasts), some of which have chemical affinities to the old rhyolite dome nearby, whilst some suggest that other rhyolite sources lie buried.



Well the eastern fringes of where the July 2014 landslide occurred formed a convenient way up to the top in this area, but this has now gone. And the landslide has covered over more (if not all) of what was a poorly exposed 1920s basalt lava. The debris dumped onto the lovely little 1920s basalt lava of Bátshraun (0robably 1921) will have covered some of the exposures I was working on – which provide evidence of lava-ice/water interactions at the time of its eruption.

Consequence. A straightforward route up to the rim at this point has now gone. It went up the eastern edge of what came down in July 2014.

Figure J. Consequence. A straightforward route up to the rim at this point has now gone. It went up the eastern edge of what came down in July 2014.

Photo taken August 2010 shows a basalt lava flow from the 1920s which is now largely/wholly covered by debris from the July 2014 landslide.

Figure K. Consequence. Photo taken August 2010 shows a basalt lava flow from the 1920s which is now largely/wholly covered by debris from the July 2014 landslide.

Flow front of the 1920s Bátshraun basalt lava, showing typical a'a upper surface (to left) with glassy and block-jointed lava at lake level indicating more rapid cooling of the loer part of the lava flow.

Figure L. Consequence. Flow front of the 1920s Bátshraun basalt lava, showing typical a’a upper surface (to left) with glassy and block-jointed lava at lake level indicating more rapid cooling of the lower part of the lava flow. See Figure F for location (debris-covered lava).

Figure M. Detail of blocky and glassy texture of Bátshraun basalt lava, showing pseudopillow fractures (long and curving with small joints perpendicular to main fracture). On right is actual pseudopillow fracture surface.

Consequence. Figure M. Detail of blocky and glassy texture of Bátshraun basalt lava, showing pseudopillow fractures (long and curving with small joints perpendicular to main fracture). On right is actual pseudopillow fracture surface.

Working in this area one is aware of the regular small rockfalls from the steep north-facing wall of the 1875 AD caldera, and of the larger slumps that have taken place. As mentioned above I was surprised at the rapid displacement of lake water that let to such a dramatic tsunami wave being formed, but then I’m a volcanologist and not a landslide expert.

No doubt landslide experts will evaluate the potential for additional landslides from the zones adjacent to the July 2014 headwall, as these may have been weakened and potentially be ready to go. However these zones appear fairly small in comparison to the estimated c.800-900 m length of caldera wall that collapsed on 23/24 July.

The Future?

There will be further landslides at Askja simply because the 1875 caldera is still ‘settling’ and will be for some time, with the southern and eastern caldera walls being likely sources because this is the area which underwent the largest amount of subsidence as a consequence of caldera formation (i.e. a sizeable chunk of pre-existing elevated terrain disappeared from the SE corner into the developing caldera). The southern walls of the caldera are particularly steep and consequently material shed from this area has a high probability of entering the lake and displacing water.

An interesting research project would be to look specifically for evidence of past tsunamis at Askja lake, to evaluate whether the July 2014 event was an extreme/low probability event, or just the latest in a number of larger events. My hunch (based purely on the pristine surface of the Bátshraun lava flow prior to this event that is now covered in debris – see Figure F) is that these larger events are infrequent.

The spectacular large landslide of 22/23 July won’t stop me working at interesting localities along the shoreline of Askja lake in future as the risk of a repeat seems very small (though this may change if the authorities carry out a more thorough examination of the source zone and say otherwise). At present the authorities are allowing access only to the relatively safe areas well above the lake level. It will be interesting to see whether this changes over the next few weeks.

In any case, Askja is a truly spectacular place to visit even if you don’t get to go down to the lake edge. And its dynamic nature has been superbly illustrated by this recent landslide, along with its effects and aftermath.


Posted August 2, 2014 by davemcgarvie in Volcanism

Tagged with , , , , , ,

The Spider Stone of Dalmellington, Scotland   2 comments


A short walk from the Ayrshire town of Dalmellington in southern Scotland is an old driveway into what was once the Camlarg Estate. Much of this driveway is still flanked by enough boulders to suggest that the original driveway was flanked by two continuous files of silent rocky sentinels.

However there is one stone that is markedly different from the rest – this is The Spider Stone, and why it’s called this will become apparent soon.

Nothing reliable is currently known about why The Spider Stone came to be where it is now, or how it got its web-like pattern. Encouraged by my old friend John Paterson who’s lived in the area for 57 years, I went to have a wee look after being told “you know about rocks Davie, come and have a look”.

What follows are some preliminary and relaxed observations made over the course of twenty minutes on a pleasant early summer evening. I’ll keep terminology out of this account as much as possible, but when you see [PN with some text] – this stands for ‘pedant note’ and will keep those with greater geological expertise slightly happier.

Caveat. I spend all my time looking at volcanic rocks, so my sedimentary work is rusty to say the least. The observations below should be OK, but don’t take the interpretations too seriously as more work (especially lab work) is needed to validate these.

I’ll put the more interesting stuff at the start (Overview, what I think, and some hypotheses), with more information at the end on rock types and further notes etc.


John Paterson of Dalmellington beside The Spider Stone.

John Paterson of Dalmellington beside The Spider Stone.


The Spider Stone lies with its web-like face upwards, but John reminded me that when we were boys it was more upright (about 30 degrees off vertical). The stone is roughly 1.6-2 metres in diameter. The Spider Stone is formed of two units of sedimentary rock – a thick lower unit and a thinner upper unit (more details are at the back). The web-like pattern is formed in the upper unit. The grooves forming the web-like pattern extend 1-6 cm into the upper unit, with most being in the 1-2 cm range. The grooves were not observed penetrating more than 1 cm into the lower unit.


Grooves on The Spider Stone, showing they are only 1-6 cm deep.

Grooves on The Spider Stone, showing they are only 1-6 cm deep. Pen is 15 cm long.

The Spider’s Web

Actually it’s not a very good spider’s web, as those constructed by spiders have long and continuous radiating spokes whereas those on The Spider Stone are discontinuous. But hey – it’s a great name and I can’t think of a better one.


The spokes and elliptical grooves of The Spider Stone. Pen is 15 cm long.


Observations (the web)

  1. The web pattern is not circular – it is elliptical.
  2. The ellipse is not symmetrical, it is skewed (bulges) to the top left looking up the stone from the base.
  3. The ‘spokes’ are discontinuous, although the segments of the ‘spokes’ are generally within a cm or two of each other.
  4. The elliptical grooves dominate the structure.
  5. The elliptical grooves are clearly more continuous than the spokes, but still appear discontinuous because no single groove can be traced through 360o without making a slight step to enable continuation. (This could easily be investigated further by a small and detailed study.)
  6. There appears to be no difference in depth between the spokes and the elliptical grooves. (Another prime candidate for a small-scale detailed study.)
  7. Some of the grooves forming the web pattern extend beyond the web into the surrounding rock.
  8. Many of the elliptical grooves terminate by becoming shallower (pinching-out).
  9. The grooves form ‘blocks’ and the shapes of these vary in a regular fashion.
  10. In general, the blocks at the centre of the web have side lengths that are broadly similar (within a few cm) whereas those at the margins have long sides that are markedly larger than the short sides. [PN shapes are more equidimensional towards the centre.]
  11. The grooves are more ‘U’ shaped than ‘V’ shaped. And in the bottom of many grooves is a white crystalline mineral that forms a flattish base to the groove.
  12. We didn’t have a decent bit of steel to test the hardness of this mineral. (Quartz is harder than steel, so a metallic smear will be left on the quartz if you try scratching it with steel; calcite is softer than steel so the steel will score a scratch in the calcite. But it could be one of the many white minerals that exists, such as barite.)
Whitish mineral occupying base of grooves. Pen nib for scale.

Whitish mineral occupying base of grooves. Pen nib for scale.

What do I think about The Spider Stone?

  • The stone is not a lava, it is sedimentary in origin.
  • The web pattern is developed in a specific ‘layer’ of rock that is (subtly) different to the rock type forming the bulk of the stone.
  • The web pattern is a surprisingly mix of regular and irregular grooves. None of the grooves appears to be continuous.
  • How did it get the web pattern? At the moment I favour a semi-natural origin modified by human action. I think the grooves were largely/partly formed by natural processes, and were enhanced/extended by people. The reasons for this are within the hypotheses below.


What am I going to do next?

Make a visit to carefully sample some of the whitish mineral to determine its precise chemical composition using some expensive equipment. Take small samples of the two rock units (away from the web itself) to characterise these properly.


Concluding note

No need to read further. But if you want to see the preliminary hypotheses and the descriptions of the two rock units, feel free to read on.



Geologists construct hypotheses based on preliminary observations to focus their thinking, and then undertake further research to prove/disprove them, or to modify them, or to formulate completely new ones.

Popular thinking about The Spider Stone considers two possibilities: it’s either natural or people carved it. From this I’ll construct four hypotheses, with some comments on how likely they are, and with notes on how to prove/disprove them. Although there are many more hypotheses than these, I’ll stick to those that are at least partly supported by my preliminary observations.

  1. One or more people decided to carve a web-like pattern into the surface of a large sedimentary boulder. To enhance the web pattern they poured a mineral solution into the grooves to form the whitish mineral. Problem – why not make the spokes and the elliptical grooves continuous and spaced more regularly?
  2. One or more people decided to carve a web-like pattern into the surface of a large sedimentary boulder, and simply deepened some existing grooves (formed by weathering, erosion, dissolution etc), with/without creating new grooves to complete their pattern. To enhance the web pattern they poured a mineral solution into the grooves to form the whitish mineral. Problem – the know-how to create a suitable mineral solution from which the whitish mineral would precipitate and crystallise.
  3. Entirely natural. The web pattern is simply a product of weathering, erosion, dissolution etc. The whitish mineral occupying the grooves was introduced during a separate event in which the stone was buried and fluids were circulating in the crust and then precipitated in the grooves. Problem – requires two separate geological processes, which is a big ask, bus in not impossible in an area with dynamic earth movements (i.e. there are many known faults in the nearby coalfield, and we are not far away from the big Southern Uplands fault zone).
  4. Entirely natural. The web pattern is simply a product of weathering, erosion, dissolution etc. The origin of the whitish mineral occupying the grooves is from the dissolution of the lime-rich upper surface that produced a purer calcium carbonate liquid which then crystallised in the grooves. Problem – I don’t know if this can actually happen, and in any case it assumes that the upper later is more lime rich and that the whitish mineral is calcite (as yet unproven).

At the moment I slightly favour hypothesis 2. One important key to unlocking the origin of the web pattern is the nature of the whitish mineral occupying the grooves. If this is something that could easily be produced by people then this argues for the grooves being of natural origin but modified by people, and then being filled with a solution that precipitated the whitish crystals. If this mineral could only form via geological processes (i.e. requiring specific pressure and/or temperature conditions outwith the human ability of the time) then the entirely natural hypotheses 3 and 4 come into play.


Rock types [PN lithologies]

Lower unit. This is a fragmental rock that displays distinct layering. Hand lens examination suggests fine-medium sand. The distinct layering is on the sub-mm scale and individual layers [PN laminae] cannot be traced across the entire exposed surface [PN so it’s not planar lamination]. The layers have numerous shallow-angle cuspate/lenticular structures. Preliminary interpretation: This is a fine-medium sandstone with a planar structure containing numerous small-scale lenticular structures. [PN. Likely to be lenticular and/or flaser bedding.]


Upper unit. This is a fragmental rock with fine-scale layering at the base that becomes less well defined towards the uppermost surface. The grain size is slightly larger than that of the lower unit, and medium sand is suggested. In places the boundary between the lower and upper units is indistinct, but elsewhere it is marked by a distinct groove presumably reflecting an impersistent joint/fracture plane at the boundary. The uppermost part of this unit has a finer-grained appearance, but lack of a fresh surface prevented further examination. (Next time I’ll bring a wire brush to gently remove the surface crud.) Preliminary interpretation: A medium sandstone representing a separate packet of sediment to the lower unit. The larger clasts at the base grading upwards to smaller clast would suggest [PN assuming present orientation is same as original] that larger clasts settled first followed by successively smaller clasts. [PN graded bedding.] A tentative suggestion based on the finer-grained nature of the top 1-2 cm along with the weathering pattern on the exposed edges  is that the uppermost portion of the upper unit is more lime rich. [PN it’s more calcareous, and the matrix especially may be more calcareous.]


Edge of The Spider Stone, showing lower unit with distinct laminations and upper unit showing less distinct laminations with coarser clasts.

Edge of The Spider Stone, showing lower unit with distinct laminations and upper unit showing less distinct laminations with coarser clasts.


Further Notes

What is known about The Spider Stone? Very little. Even my friends who have lived in the area for over 50 years know nothing definitive, just vague rumours. The Camlarg Estate had ceased to be long before this, and so reliable local knowledge on the origins of Spider Stone probably disappeared long ago. There may be older folks in Dalmellington who can recall knowledge passed down to them, but they would need tracking down and interviewing.

A web search gives no images, and only one decent link to a Geocache site. This gives no references to the description provided, and no observations to support the interpretation.


The cache is located a short distance beyond The Spider Stone. It is thought that the Web shaped fissures in the face of the rock were created by gas bubbling up to make a mound and then cooling. It is found in volcanic conditions, probably in the lava flow. There are a number of rocks along the former driveway to the old house which are thought to have been imported in Victorian times. We are doing extra research on this stone in collaboration with the local community. Locals have been known to make wishes at the stone. A rare geological gem.congratulations to stewart57 on FTF

The Charm of Columns   2 comments

Devil’s Postpile (US), Fingal’s Cave (UK), Giant’s Causeway (Ireland) – these are nature’s beauties because of their spectacular columns. But how do columns form? Well this illustrated and short blog post is a selective stravaig through the charming world of columns, with a focus on lava-ice and lava-water interactions. It also provides a bit of insight into the science behind column formation, as well as hinting at unexplored areas and discoveries yet to be made.


The massive interiors of lava flows often show different zones, and where a zone consists of well-developed columns it is given the term ‘colonnade‘, and the simplest of lavas has an upper colonnade and a lower colonnade which meet (usually imperfectly) towards the  middle of the flow. ‘Entablature‘ is the name given to a zone displaying irregular columns oriented in various directions, or a zone comprising much smaller blocks (known as ‘cub-jointed’ lava or ‘kubbaberg’). Entablature is interpreted by all workers as representing penetration of additional coolant (water/steam) into a lava flow, and thus is a useful environmental indicator.


When a basalt lava flows down an Icelandic river valley. Lower colonnade demonstrates that water did not interact with the base of the flow, but the upper entablature zone indicates that river water flowed across the top of the flow and down into the still hot lava.


One of the plus points of being an academic and having a bit of time to do research is that you can get a good PhD student working on something that really fascinates you. For many years I have looked at columns and other fractures in lavas, especially those involving volcano-ice interactions, and I knew there was some good science to do on them. So I managed to persuade a couple of colleagues that a PhD project entitled ‘Lavas in Glacial Settings’ would be a winner, and of course you cannot study these lavas without being surrounded by columns. So part of the project became an in-depth look at columns and fracture formation in lavas, and we had some nice surprises and made some new discoveries.


In the land of rhyolite columns. Entire rhyolite domes hundreds if metres high have columns throughout, indicating effective penetration of coolant into the interiors of domes and their constituent lobes. Iceland.


A dacite lava lobe that flowed in an ice tunnel – this is the side of the lava lobe where it was moulded against the ice wall. Villarrica volcano in background. Chile.


Way-up indicator! The pipes form vertically as they propagate upwards. So the columns formed in an inclined orientation and afterwards. Basalt lava, Iceland.

I am going to show some examples of fractures that illustrate a number of points. But first, some basics about columns. For those who want further information I have added ‘geek notes’ but you really don’t have to read these to understand and enjoy the images.


Small-medium columns in a basalt intrusion into a moist fragmental host (volcaniclastics). Iceland.

The basics

  1. Columns form due to thermal contraction of cooling lava.
  2. They form at right angles to the cooling surface, so a horizontal lava body (e.g. lava flow or sill) will have vertical columns, whereas a vertical intrusion (e.g. dyke) will have horizontal columns. [Geek note: Columns propagate in the direction of the thermal gradient defined by the isotherms in the cooling lava body. In a horizontal lava body the isotherms will be horizontal, and therefore the columns will propagate vertically.]


    Lava flows with vertical columns at their base, forming a ‘colonnade’. The upper zone of more irregular columns oriented in various directions is ‘entablature’. Iceland. Look carefully at the basal colonnade and you will see that the early formed columns are small (rapid cooling) and that they merge upwards to form larger columns reflecting a more stable and slower cooling regime.

  3. Column diameter is related to cooling rate, so for example a faster cooling rate will produce smaller columns. [Geek note: Column diameter is controlled by the viscoelastic response of lava to cooling, so faster cooling leads to smaller-diameter columns. There is as yet no widely-accepted model which relates column diameter to cooling rate. There’s a PhD project for you…!]

    Large vertical columns in a basalt lava flow in west Iceland. Note the horizontal ‘chisel marks’.


    Small columns in a rapidly-cooled rhyolite lava. Rule is 1 metre. Iceland.

  4. Columns form in lavas of all compositions, though basalt columns are best studied. [Geek note: Rheological factors mean that basalt columns best approach the ‘equilibrium’ condition which is the formation of hexagonal columns with equal sides. Rhyolite columns get nowhere near this ‘equilibrium’ condition.]

    Large rhyolite columns at the top of a subglacial dome.


    Rhyolite columns around 0.5 metres in diameter. How many hexagons with equal sides can you identify?

  5. Columns do not form smoothly – they form in distinct steps. This process is reflected in the ‘steps’ that can be seen on many columns, and on vertical columns these ‘steps’ are horizontal. These have various names, such as  ‘chisel marks’, ‘striae’, ‘chatter marks’ and ‘step-wise advance cracks’. I’ll call them ‘chisel marks’ in this blog.  [Geek note: These form parallel to the long column axis. They are beautiful examples of cyclic fracturing in a uniform stress environment where fracture propagation is retarded until further cooling and contraction enables the tensile strength of the lava to be overcome. One it has been overcome the fracture is initiated and propagates from cool into hotter lava, but stops as the fracture encounters lava too hot to fracture in a brittle manner. Then the process starts again.]


    Chisel marks (horizontal) on near-vertical columns, with inclined flow banding, in a rhyolite lava. Iceland.

  6. Small but subtle features at the chisel marks enable the direction of formation of the columns to be determined. [Geek note: these ‘hackle’ marks and plumose structures identify the point of fracture initiation in the cooler part of the lava and the direction of fracture propagation into the warmer part of the lava.]


    The little ‘hackles’ starting at the chisel mark at ’22’ on the left side of the rule and terminating upwards, indicate that cooling was from the base upwards. Dacite lava, Chile.

  7. In a horizontal lava body such as a lava flow or sill, cooling occurs both at the top and that the bottom, so columns grow upwards from the bottom and downwards from the top. They meet in the middle, often surprisingly well, though an accommodation zone of imperfect joins is usually present. [Geek note: The upper columns are usually longer than the basal columns, suggesting that cooling from above is more effective than cooling from below. This is not surprising, as convective removal of heat from the top is generally more efficient than conductive removal of heat from the base.]
  8. Not all lava bodies have columns, and the reason why they don’t is probably another PhD project….

Upper and lower colonnades with well-formed but short vertical columns. Interior zone comprises most of the flow and consists of entablature, which I informally call ‘chevron’ entablature. This is interpreted as a lava flow which dammed a river, and whilst the river was rising behind the dam the lava was cooling from above and below. When the dam broke the released water flowed across the lava top, percolated down into the flow, and the more rapid cooling produced the thick entablature zone. Iceland.


The broader context of the previous image. This lava flow, c.6,500 years old, has flowed into a river valley. The edge of the flow is exposed left of the waterfall, and the lava thickens to the right of the waterfall into the palaeovalley. Entablature is not present at the edge of the flow – it is only present in the thicker and lower-lying parts of the flow.


Lower colonnade with upper entablature zone, with this entablature zone being of what I informally call ‘cube-jointed’ entablature (equivalent to kubbaberg). Iceland.


The transition zone where an upward growing lower colonnade meets cube-jointed entablature propagating downwards. Iceland.


The sub-horizontal columns on the right-hand side of this glassy dacite lava in Chile indicate a vertical cooling surface to the right. Given the the lava is thin on the ridge crest (top left) and thickens down to the right, the interpretation is that when the lava flowed off the ridge crest it encountered ice.



The author examining a remarkably smooth surface within a subglacial rhyolite lobe. Iceland.

Posted May 18, 2013 by davemcgarvie in Uncategorized

Katla’s next eruption?   8 comments


Just when a volcano seems to have settled into a ‘pattern’ it starts misbehaving. Welcome to Katla.

The ‘pattern’ involves 16 recorded Katla eruptions over the past c.830 years, with the last 9 being the most reliable as the actual months in which eruptions took place were recorded. And from this comes the (admittedly rather bold) statement that “Katla erupts twice a century, and the last nine eruptions have all been in the May-November period “. [Not my statement by the way….]

But Katla has not had a sizeable eruption since 1918, and the current repose period of 94 years is the longest known since reliable records began. [Prior to the present one the longest is considered to be the c.80 years between the (approximately dated) eruption of c.1500 and the August 1580 eruption.]

Most geologists who study volcanoes in Iceland and who are aware of Katla’s eruptive history think it’s reasonable to expect that Katla’s next eruption is likely to be similar to her last sizeable basaltic eruption in 1918. But three additional factors may have an unknown influence on the next eruption: (1) Eyjafjallajökull’s eruptions in 2010; (2) probable small subglacial eruptions in 1955, 1999, and 2011; and (3) that an emergent Katla eruption is ‘overdue’ according to a pattern that has persisted since c.1180 AD. (Emergent is a term volcanologists used to describe when a subglacial eruption breaks through the overlying ice and produces a distinct atmospheric eruption plume.)

To add further complexity, it’s worth noting that Katla doesn’t just produce sizeable basaltic eruptions within her ice-covered caldera, so she could surprise us.


Katla from the South, taken in 2012. A broad dome of ice up to c.700 m thick within which sits the Katla caldera.

Katla’s 7 types of Holocene eruptions

So let’s start by briefly describing listing the main types of Katla eruptions, with a few comments for me. And to keep it simple I’m going to restrict myself to the Holocene (i.e. the period from the end of the last glacial period till the present day, c.9,000 years), as Katla’s older geological history is largely unknown.

1. Small subglacial basaltic eruptions that have not become emergent (i.e. they haven’t pierced the overlying ice). These produce modest glacial outburst floods (jökulhlaups), and at the eruption site the overlying ice collapses to form a distinct depression (cauldron). Examples include the 1955 event, and the July 2011 event. (The word ‘event’ is used deliberately as the lack of prima facie evidence of actual eruptive products means that the heat source required to melt the ice could be geothermal rather than magmatic.)

2. Modest sub-ice to emergent basaltic eruptions. These produce modest-large jökulhlaups, and modest tephra falls. There are many examples in the historic period (i.e. last c.1100 years since the Viking settlement), including 1823, 1860, and 1612.

3. Larger basaltic eruptions that swiftly go emergent. These produce the biggest jökulhlaups, accompanied by thick falls of tephra (especially in proximal areas). Examples include the most recent sizeable eruption in 1918, and the huge eruptions of 1721 and 1755 and 1625. One interesting enigma is how these eruptions produce so much meltwater when so much thermal energy is lost to the atmosphere, especially when there is also no obvious place to store substantial volumes of meltwater at Katla. (A nice problem waiting to be solved.)

4. Modest eruptions of silicic magma. Yes, Katla also erupts silicic magma (dacite to trachydacite, and even some rhyolite). The evidence of these eruptions comes from tephra layers in the surrounding soil profiles, which have the lovely name of SILK layers. If they produced any effusive products (domes, lava flows), these either lie buried beneath the ice or have been removed by erosion. Interestingly, there have been no eruptions of this type since the massive Eldgjá fissure eruption in c.934-938 AD (described under 7 below). At least 12 SILK eruptions are known between c.6600 and 1685 years ago.

5. Large eruptions of silicic magma. Only one is known with certainty and I’m stretching beyond the Holocene to include this as it occurred c.12,000 years ago. It produced a sizeable Plinian to sub-Plinian eruption plume, and more interestingly from the hazard perspective, a number of pyroclastic flows (known as ‘pyroclastic density currents’ to the volcanology pedants). The deposits from the pyroclastic flows form the Sólheimar ignimbrite, which is exposed on the ice-free flanks of the volcano. Parts of this eruption are rhyolitic.

6. Small-modest basaltic eruptions in the fissure swarm associated with Katla (which trends to the northeast). These are neither well studied nor well exposed, largely because they are either covered by younger and more voluminous lavas, and/or because they have been eroded.

7. Large basaltic eruptions in the fissure swarm associated with Katla. Two examples are known – the Hólmsá Fires which erupted c.6600 years ago, and the Eldgjá Fires which erupted c.934-938 AD. An interesting oddity is that unlike the other volcanoes in this part of Iceland, Katla proudly displays a prominent fissure swarm. From this fissure swarm emerged Iceland’s largest historic basalt eruption – Eldgjá. Surprised? You might be because Laki  gets all the attention, largely because it happened more recently in 1783-85, and also because its effects on the Icelandic population, and on the population of western Europe, and on the climate, are much better known. The Eldgjá eruption is estimated to have vented c.18 km3 of lava, during an eruption that lasted up to 5 years.


From above the town of Vík, looking East. The black sand is derived from Katla eruptions, and is deposited by jökulhlaups (glacial outburst floods). After the 1918 eruption, the new land formed by the jökulhlaup deposits was Iceland’s southernmost point (Kötlutangi), now eroded back nearly to the ‘island’ of Hjörleifshöfði.

Katla’s 3 eruption groups

Let’s simplify and put these eruption types into three groups:

Group A ( types 1-3) are basalt eruptions within the Katla caldera.

Group B (types 4 and 5) are silicic eruptions (rhyolite-dacite-trachydacite) within the Katla caldera.

Group C (types 6 and 7) are basaltic eruptions in the fissure swarm.

From this we can formulate two simple conclusions:

  • From the ice-covered Katla volcano itself (defined by the caldera), two kinds of eruptions are known to occur – basaltic eruptions and silicic eruptions. These vary in volume and explosivity. However, basalt is volumetrically the dominant composition, and is also the most frequent composition erupted.
  • The biggest eruptions occur in the fissure swarm to the northeast, and only basaltic eruptions are known from there.

Katla from the South, showing some of the older (unstudied) Pleistocene rocks forming the lower flanks, with some paler coloured silicic rocks exposed in places.

So what’s happened to the so-called ‘predictable’ pattern?

Ah, yes, the curious case of the misbehaving volcano. Well, the Viking settlement of Iceland started in 874 AD and then 60 years later the massive Eldgjá fissure eruption. And being reasonable-to-good recorders of events, especially ones that affected the populated and well-travelled coastal strip, the new settlers and their descendents wrote down some useful information on Katla eruptions – especially the ones that caused floods as these travelled over the well-travelled paths and tracks on the coastal strip on their way to the sea. From this archive comes the ‘predictable’ pattern that “Katla erupts twice a century, and the last nine eruptions have all been in the May-November period “.

This ‘twice a century’ pattern has been surprisingly consistent for the last 16 recorded Katla eruptions over the past c.830 years

However within this persistent pattern there have been some remarkable fluctuations in how much erupted and the repose periods between eruptions. I’ll mention one that really impresses me, which is that two of Katla’s largest eruptions of the past c.1100 years occurred in 1721 and 1755, just 34 years apart. I think it is rather splendid that a volcano can regularly erupt twice a century for eight centuries and only in the May-November period, whilst at the same time having zero correlation between how much is erupted and the repose time between successive eruptions. Think about it.

The ‘breaking’ of the predictable pattern is that Katla has not had a sizeable eruption since 1918. So at time of writing (2012) it’s a repose period of 94 years, which is the longest recorded.


The spectacular ice cauldron produced in the surface of the ice after the probable 1955 eruption, which if it was an eruption remained entirely within the ice. Photo taken by Sigurður Thórarinsson.

A prediction?

If I was a betting man my prediction is that the next sizeable eruption at Katla would be a basaltic eruption within the Katla caldera, one that is likely to be emergent but may be larger or smaller than the 1918 eruption, and that will occur within the next century. (Not much of a prediction you might say, but it is reasonable and does fit the data.)

But how likely is it that something different will happen?

Well here’s three possible scenarios based on the three groupings above (A-C), along with a few comments.

Possible eruption scenarios

Group A (basalt eruption from Katla). If it can be assumed that the 1955 wholly sub-ice eruption may have eased the pressure on the magma plumbing system (which I admit is speculation on my part), then this, along with another pressure-reducing event in July 2011, may be the reason why Katla’s long-expected sizeable 1918-style eruption is lagging behind schedule.

The July 2011 event also heralded a surprisingly and rapid change in the pattern of seismic unrest in the area, from a high level of unrest in the west (in the Goðabunga area beneath the ice and slightly outwith the caldera rim), to a high level of unrest further east focused within the Katla caldera proper. If this change in the location of seismic unrest is due to a switch in where the heat is rising through the crust (and that’s not certain) then things are heating-up within Katla herself.

Group B (silicic eruption from Katla). How silicic magma is produced and stored within Katla is not well understood. However since the massive Eldgjá eruption there hasn’t been a silicic eruption from Katla, and some scientists consider this to reflect that the magma plumbing system beneath Katla was changed. It’s a fair conclusion, and a high throughput of basalt through a volcano’s plumbing system does not provide optimal conditions for the production and storage of silicic melt. Look at Grímsvötn for example – the most frequently erupting volcano and not a sniff of rhyolite anywhere (at least not that we know of). Anyhow, I would be quite surprised if Katla’s next eruption was a silicic one, but I’d also be quite delighted as it’s my favourite magma.

Group C (basalt eruption in the fissure swarm). If this is of the scale of Eldgjá, it’s the nightmare scenario for Iceland. And it could also be a nightmare for western Europe also (remember Laki 1783). However if it follows a similar pattern (especially eruption rate, duration, meteorological conditions) as Eldgjá then it reduces from nightmare to ‘wow what a big and amazing eruption’. Laki was so damaging to Iceland and western Europe because a lot of magma and associated volatiles (especially S, F, and Cl) were vented in a short time. Eldgjá however, vented more magma with similar volatile content – but over a longer period. This is a rather critical point, because the longer the duration of an eruption the less nasty the effects of the volatiles because additional dispersal time means that their concentrations diminish, especially in distal areas like western Europe. Upwind of such an eruption (and orthogonal to the fissure direction) would be a tourist’s paradise – an opportunity to witness what some volcanologists would call a flood basalt eruption. Call it what you will, it would be truly spectacular. And to watch the development of a massive lava flow field over a few years – a volcanologist’s dream. Anyhow I’d be most surprised if Katla produced a large Eldgjá-sized eruption as its next event, but I’d be equally delighted if I was still around to witness it.


Black beach formed of jökulhlaup deposits and pulverised basalt from nearby old and unstudied formations that may be connected with Pleistocene eruptions of Katla. The promontory in the background is Dyrhólaey, which is Iceland’s southernmost ‘solid’ part of the mainland, and which may also have been produced by Katla during the Pleistocene.

Concluding points

I’ll finish on two Iceland-specific points. The first is that the Icelanders know more about Katla than anybody else, as they have been living with her for c.1100 years and modern-day Icelandic scientists have been monitoring her with considerable intensity and with increasingly sophisticated equipment. And that’s why it was brilliant when Eyjafjallajökull erupted twice in 2010, as all this kit was already in place and teaching us a huge amount about Eyjafjallajökull’s plumbing system. Anyone writing a professional paper on Katla would be well advised to collaborate with the Katla experts in Iceland. This blog entry allows me to synthesise, speculate, and muse in my own individualistic fashion.

It is also worth emphasising that the Icelanders have plans on how to protect people when Katla erupts, and if you wanted evidence of how thorough and good these plans are, just reflect on what happened when Eyjafjallajökull erupted twice in 2010 and how wonderfully effective and safe the evacuation plans were.

The second point is that had this been a peer-reviewed paper I would have cited my sources of information throughout this account. But again it’s a blog so to keep the text flowing I didn’t. Nevertheless I have sifted and synthesised information from a number of sources and a few of these are listed below.

Finally, for the sake of one of my best friends in Iceland I rather hope that Katla erupts sooner rather than later. You see she is also called Katla, and every time there is a news report speculating about when Katla is going to erupt she gets teased by her friends with “come on Katla, stop keeping us guessing – tell us what you are going to do”. Ah the perils of naming children after volcanoes. At least she wasn’t called Eyjafjallajökull or Ok (yes, Ok is a volcano in Iceland).


Katla from the East.


A useful Icelandic web source is http://earthice.hi.is/katla_bibliography

Peer-reviewed papers written by the scientists listed below (in no particular order) contain reliable and credible information on Katla, and can be accessed via Google Scholar etc. Note that this is not an exhaustive list: Guðrun Larsen; Bergrún Arna Óladóttir; Magnús Túmi Guðmundsson; Helgi Björnsson; Finnur Pálsson; Thorvaldur Thordarson; Brindís Brandsdóttir; Christian Lacasse; Olgeir Sigmarsson

Posted November 6, 2012 by davemcgarvie in Uncategorized

Tagged with , , ,

Alaska – The Valley of Ten Thousand Smokes (Part 1)   2 comments

Exactly 100 years after the start of the largest eruption of the 20th century I walked into the site of the eruption – The Valley of Ten Thousand Smokes. Had I been there a century earlier I would have been walking into the Valley of Death. So let me tell you a little of why I was there and what I did.

Float plane. The fun and exciting way to arrive at Brooks Camp.

I was participating on an international volcanological field camp that takes place annually, led by John Eichelberger (in charge of USGS volcanic hazards program) and Pavel Izbekov (a perpetually smiling volcanologist originally from Russia who does some crazy stuff on active volcanoes).

So the start of the trip involved a float plane ride from King Salmon to Brooks Camp on the edge of the great Katmai wilderness. This was my first trip in a float plane and it was exciting. For any aircraft geeks I went out in a Beaver and came back in an Otter (classic De Havilland aircraft). A story is told of a pilot who when he first landed at King Salmon thought how friendly the locals were as they were all waving at him. When he opened the plane door and the bugs attacked he realised why they were waving. Yes, the bugs are plentiful and aggressive and they acquired a real taste for Scottish skin.

Tundra – an undernourished young female brown bear.

At Brooks Camp there are cabins and bears. In fact the bears often wander between the cabins, and when I was there we were told to be a little extra careful as the girlfriend of a mating pair kept hiding from her mate, and so the large boyfriend was often found wandering around the cabins disconsolately looking for her. It was early in the season (i.e. the salmon runs that attract all the bears had not yet started), so I only got a good view of one rather underfed-looking female. The park rangers have been told not to give names to bears (avoids anthropomorphising them apparently), so I was told this was bear 130. And quietly in my ear the doe-eyed female ranger whispered “we call her Tundra, but don’t tell anyone”. Despite the abundance of bears and their close proximity to humans there has never been a bear attack. We did of course get taken straight to a bear briefing the second we got off the float plane, and had to pass a quiz to get a badge and be allowed out. Perhaps a good example of the power of education combined with such a place tending to attract those who have a respect for wildlife and the wild places? But I digress.

First glimpse of the pyroclastic flow deposits (basically an ignimbrite sheet) from the bus journey in. The distal ends of The Valley of Ten Thousand Smokes appears as the pale brown material entering right of centre.

After a bus ride to the trailhead, we started our hike into The Valley of Ten Thousand Smokes (VTTS for short). It was going to be 9 miles and c.2000 feet of ascent, and it usually takes around 8 hours. But this part of Alaska had received what locals call their ‘hundred year snowstorm’ and this dumped a huge amount of snow late in the season. The effects of this we were to discover on the hike in to our base  – the Baked Mountain huts.

After crossing Windy Creek. Edge of pyroclastic flow and fall deposits in the Valley of Ten Thousand Smokes.

It had been a while since I had carried a heavy pack carrying all my field gear and food for 12 days, and the scales at Brooks Lodge told me that I was lumping 83 pounds on my back. There were many times on the walk that I wish I’d left the whisky behind, as this would have saved 3 pounds in weight. The day started warm and sunny, and most of us were in shorts, which was just as well as a fast-flowing creek had to be crossed (in sandals), and the lighter participants helped across. In fact one had to be held between two of us and apparently she moved like a cat-flap. Further up the valley we stopped a few times for geological stuff and spotted a sow and two bear cubs, who scarpered after a long look at the group.

Then it got cloudy and a bit colder. And we had some difficulty negotiating gullies filled with snow. From then on we were on snow most of the time, which was a bit slushy but only knee-deep in parts. Then, with two-thirds of the hike done, the trail turned into a real scorpion. The innocuous flat-lying snowy wasteland that separated us from the slopes to Baked Mountain contained two nasty little streams that we just waded across (knee deep) as boots were already soaked. But they were pleasant compared to the nightmarish plugging through mostly thigh-deep slush, which sapped strength and chilled us to the bone. Many of us who fell over had to slip off our packs, crawl till we could stand again, then lift the dripping packs back onto our backs (often with help from others) and continue. To everyone’s credit they kept going towards the huts and safety and warmth. Eventually we all made it, with some participants close to or at the limits of their endurance. There were some helpful heroics from volunteers who went back down the hill to help those who were struggling – real teamwork. It had taken around 12 hours to walk 9 miles, with the last 3 miles taking almost half of that time. One of the toughest walk-ins I have done in my career. Sorry – no photos of the epic part of this hike – all energy was saved for walking!

[Reflective note: I would not wish the above account to put anyone off participating in the future on such an amazing trip, so I’d like to stress that even though these were unusual thick-snow conditions, safety was taken care off. For example there were radios distributed throughout the group (which were used extensively during the walk-in), and had there been a problem tents would have been erected (we had three with us) on one of the solid ground refuges and stoves fired up to cook food and replenish energy, and also to provide warmth.]

The Baked Mountain Huts. Water from melted snow. Structure on the right is the ‘outhouse’ or ‘rest-room’ in US parlance. It’s of the long-drop non-flushing variety.

The Baked Mountain huts comprise a hut for 6 (which is leaky and mouldy) and a ‘VIP’ hut for 4 to which I was invited. But when I discovered that one of the VIPs was a legendary snorer I decided to look elsewhere.  So I cleared a space in the floor of the storage hut used by the tent people, which I had to vacate early every morning, thus depriving me of my usual slothful enterprises whilst in the field.

On the slopes of Baked Mountain, examining pumice from the 1912 eruption that forms a thick blanket over the mountain. The Valley of Ten Thousand Smokes is behind, which now forms a flat-lying plain (incised in places) of pyroclastic flow and fall deposits, which reach up to 300 m in thickness.

So that’s the lengthy preamble – now for the spectacular scenery and geology.

Griggs Volcano above the clouds as we descended below the clouds, from high on Trident Volcano.

Right across from the Baked Mountain huts is Griggs Volcano, which has a number of young (Holocene) lava flows but no recent/recorded eruption. It looks rather fresh in parts, so is just sleeping. And from the top of Baked Mountain the snow-clad volcanoes of Magiek and Martin can be seen. These volcanoes (and others which come later) are part of the Katmai Cluster, which is a quirky group of volcanoes for which there is no good explanation (in my view) for the ‘clustering’.

Magiek Volcano. (Martin Volcano is hidden behind it.)

Magiek Volcano from Baked Mountain, with Martin Volcano just visible behind it.

But the real enigmatic part of this cluster is also the most dramatic, and its existence produced the VTTS. The small c.500 m diameter rhyolite dome of Novarupta looks innocuous, but it was the final act of a c.60 hour (3-day) eruption that was the largest of the 20th century. The dome occupies a backfilled larger vent structure which was about 3 km by 2 km in size (a very reasonable interpretation). From this vent, in 1912, roared 13 cubic kilometres of magma. That’s actual magma by the way, not the total volume of ash, pumice etc, which was much larger.

Katmai Volcano (left, arrowed) and Novarupta rhyolite dome (right, arrowed). Some 10 km apart but considered to be intimately linked during the 1912 eruption of Novarupta.

And I’m going to end Part 1 with a tantalising enigma. Actually there are two. The first is that prior to the 1912 Novarupta eruption there was no indication of a volcano in the vent area. And the second is that during the eruption a substantial mountain at the Katmai Volcano subsided to form a caldera, accompanied by only trivial amounts of magma/heat escaping during caldera formation. The hypothesis is that magma escaped from a holding chamber beneath Katmai Volcano to feed the eruption at Novarupta some 10 km away, and that the void left after magma extraction caused collapse of the Katmai Volcano. It’s a plausible story, but there are problems with it. Wait for Part 2….

Panorama of Katmai Volcano crater lake which is c.4 km in diameter which now contains c.250 m of water, and which occupies the site of what was once two-three peaks rising way above the remaining crater rim. At the time of its formation it was ‘dry’ and there was small dacitic lava dome at the bottom of the caldera, and much steam and phreatic explosions. All quite trivial in comparison to the huge energy involved in collapsing a mountain.

Posted June 23, 2012 by davemcgarvie in Uncategorized

Quetrupillán – the headless ghost   Leave a comment

There is a legend of three volcanoes in Chile. They form a line, with the tall and proud ice-capped peak of Lanín volcano at one end, and the gently smoking, demure, and shapely ice-capped volcano of Villarrica at the other end. In between the two is the stunted Quetrupillán, and which has a somewhat beheaded look about it. As if someone has removed the top of a once proud peak.


Lanín volcano (3,747 m high). The proud warrior….


Villarrica volcano. The demure and smoking lady….

The legend is that Lanín was a proud warrior who loved to gaze upon his lovely smoking lady of Villarrica. But growing between them was a warrior called Quetrupillán, and as Quetrupillán grew larger Lanín became frustrated that his view of the lovely Villarrica was becoming blocked. Frustration grew into anger and in his rage Lanín cut off the head of Quetrupillán.

Quetrupillán means ‘the headless ghost’ and this could be either a quaint story invented by indigenous tribes to explain to their children why these three volcanoes have different shapes, or perhaps it is a story passed down through time by ancestors who witnessed the eruption at Quetrupillán that destroyed the summit cone.


Quetrupillán volcano, showing the ice-filled summit crater with remnants of the original cone flanking the ice.

This is something myself and my Chilean colleague are investigating. Actually we are investigating quite a lot about Quetrupillán. But for now I’ll say a little about the trip we did back in March of this year (2012).

Quetrupillán has a famous neighbour – the smoking volcano of Villarrica. It is understandably famous because the ‘smoke’ is a noxious cocktail of magmatic gases escaping from an active lava lake that occupies the summit crater. And because it’s a fairly easy hike up to the summit to look down into the lava lake. Everyone who goes up there comes away moved if not awed by the volcano, as for most this is a truly unique and weird experience. But turn away from the summit and gaze around and the eye is drawn to the superb ice-capped volcano of Lanín, which lies on the border of Chile and Argentina. (In fact the border goes over the summit.) Lanín looks higher and in fact it is, because it sits on a separate (higher) crustal block, but this is not the place to delve into the vagaries of basement geology and discriminating between the little that is known and the large that is speculation.


Lanín in the background, with ‘beheaded’ Quetrupillán in the middle ground. Taken from the summit of Villarrica. Volcanologist colleague is wearing mask due to noxious magmatic gases escaping from the summit lava lake. Photo courtesy of Jose-Luis Palma.

Looking down from the rim of Villarrica into the lava lake. Taken in 2004, the level of the lava lake is quite low, so only occasional splashes of glowing lava were seen.

Few pay any attention to the less obvious and shorter volcano that lies between Lanín and Villarrica. But this is a rather special volcano. This is Quetrupillán.

My Chilean colleague Andres knew of my work on Icelandic volcano-ice interactions, and that I had worked on another Chilean volcano. So he asked me to go on a recce to Quetrupillán to see if there were interesting volcano-ice interactions.


The author on his trusty steed. There is no vehicle access to Quetrupillán, so the choice for those who cannot afford a helicopter is either horseback or walking. Photo courtesy of Beth McGarvie.

Riding on horseback through forests of Araucaria Araucana was rather special, as the trees exude an enduring and timeless air. The wreaths of cloud added to the feelings of otherworldliness. Photo courtesy of Beth McGarvie.

Now I’m not going to give too much away as there’s a grant proposal being written to do further research in this fabulous place! So I’m going to restrict myself to leaving you to enjoy a few images that reveal some of Quetrupillán’s beauty, and hopefully the captions accompanying the images below will say what needs to be said.


Imagination time. Imagine a lava flow creeping slowly through a circular tunnel tunnel melted into the ice. Cool down the lava flow, then remove the ice. This is what we have here. The giveaway evidence is the radial columnar joints (indicating cooling against ice walls and ice roof) and the sinuous shape of the lava flow showing that it was confined (by ice).


Laguna Azul lies south of Quetrupillán’s summit, and our camp was in the woods. The twin peaked volcano in the distance in Volcan Mocho-Choshuenco.


One for the the volcanology geeks. These top-bottom ridges on this chunk of dacite lava column represent staggered cooling of the lava, as columns don’t form smoothly – they form in increments or steps. The fracture pattern on the surface (when preserved, as here) indicates development of the fracture from solid into ductile (hot) lava, and so gives the direction in which the column was forming. In this case from left to right.

Lanín in the background, with one of the young dacite lava flows from Quetrupillán damming the outflow of Laguna Azul.


Laguna Blanca on the south side of Quetrupillán is fed by glacial streams and so has a milky appearance. In the backround is one of the young (Holocene) andesite lavas from Quetrupillán. The pale-coloured block is an example of ‘peperite’ where a lava has flowed over wet unconsolidated sediment and steam action during lava-sediment interactions has aided mixing between the two components.