This is a repost of a guest post by volcanologist Dave McGarvie, first published here in 2017. Dave is currently associated with the Lancaster Environment Centre of Lancaster University; at the time of writing this post he was working at the Open University. He studies volcanoes in Iceland and Chile and can be found at https://research.lancaster-university.uk/en/persons/dave-mcgarvie/

As a preface to this post, it is worth noting that this ignimbrite is nowadays indeed considered to be associated with the collapse of the Torfajökull caldera (as already mentioned in Dave’s post). The caldera-forming eruption is dated to either 55 thousand years ago, based on argon dating by Jonathan Moler, which would associate it with the Icelandic rhyolitic Z2 Ash, itself a component of North Atlantic Ash Zone II. (A younger date of 38 thousand years ago is derived based on zircon crystal dating, in work by Zoe Moser.) Torfajökull (the name means Torfi’s glacier) is one of the wonders of Iceland: see https://www.volcanocafe.org/the-new-lagoon-and-icelands-thor-adventure-park/ (which regrettably was about possibility rather than reality).

And now over the author:

One of the wonderful aspects of working as a volcanologist is Iceland is that fascinating new puzzles and their bigger cousins (enigmas) keep appearing. Sometimes even a fresh look at what seems to have been fully ‘sorted’ decades ago can – with the benefit of newer knowledge/understanding – result in enigmas re-emerging.

Welcome to the Thórsmörk Ignimbrite.

---

Author’s Note:

What is an Ignimbrite? An ignimbrite is simply the rock that forms from the deposition of one or more pyroclastic flows*. Something extra needs to happen to convert the loose aggregation of pyroclasts and rock fragments that comprise a pyroclastic flow deposit into a rock (ignimbrite). Fortunately, pyroclastic flows are surprising good at retaining much of their heat during transport, so once they come to a halt this heat can cause sintering and welding – and thus the formation of a rock (ignimbrite) that is much more resistant to erosion. Compaction of the pyroclastic flow deposit can also occur, and again this transforms the deposit into something more rock-like. There are a host of other post-emplacement processes that affect pyroclastic flow deposits, but we’ll leave these as they are less relevant to this particular article. In the field, ignimbrites have textures too variable to describe briefly here. Fortunately, the Thorsmörk Ignimbrite is a bit simpler and has an ash matrix (light or dark coloured) with a scattering of pale-coloured pumices and rock fragments. These reflect the components of the original pyroclastic flow – ash and pumice being fragmented magma, and rock fragments being chunks of older rocks that were fractured and pulverised during the eruption (e.g. vent walls). For simplicity, I’ll refer to the ignimbrite as either welded, part-welded, or unwelded. *Note that I have chosen to use the term ‘pyroclastic flow’ instead of the jargon term ‘Pyroclastic Density Current’ (PDC for short), as I consider PDC a cumbersome and partly-misleading term.

---

Introduction

I first looked at the Thórsmörk Ignimbrite in September 2013, and after a couple of days in the field followed by some literature reading, I realised that there were still many unanswered research questions about the Thórsmörk Ignimbrite. Here’s a few of them:
1. When was this pyroclastic deposit erupted?
2. What was the source?
3. Is there just one ignimbrite present, or are there more?
4. What was the environment like when it was deposited?
5. What processes affected the ignimbrite soon after it was deposited?
6. Why are there dark and pale phases – are they the same chemically but just different physically?
7. Why is there so much reworking, and why had previous authors not mentioned it?
8. What do fragments of welded ignimbrite within overlying unwelded ignimbrite tell us about time gaps etc?
9. What did the original complete PDC stratigraphy look like?
10. When did the processes that jumbled the ignimbrite occur – during deposition, during welding, or post-welding (or some combination of the three)?

The answers to 1 and 2 are being worked on as I write, by PhD student Jonathan Moles (http://www.open.ac.uk/people/jdm539) The provisional and exciting answer to 2 appears to be Torfajökull to the NNE, but that’s a story for another time….

Figure 1 shows the location of the various volcanoes in southern Iceland along with exposures of the Thórsmörk Ignimbrite (in green).

Nobody is currently doing serious and comprehensive studies on the Thórsmörk Ignimbrite, so many of these questions will remain unanswered for the foreseeable future.

So, the aim of this blog article is just to give you some insight into what I’ve gleaned from a few days of fieldwork, and I’ll mainly address points 4-6 above. As this is recce-level work a nice robust model can’t be provided (it would be too speculative for my liking), but I hope that you will get a bit of insight into what volcanologists do in the field, complete with uncertainties, challenges etc.

Finally, another reason for this field-based article with lots of images is that this area is very accessible, and so if you – dear reader – wished to have a look for yourself, it is straightforward to do so. I’ll be happy to provide details of specific locations. Now, onto the nitty gritty.

Variations and themes

The first thing that struck me about the Thórsmörk Ignimbrite was its variability. From the logs in papers published back in the 1980s (see list at the end) I was expecting simple horizontal or gently dipping boundaries between different flow/cooling units that could be traced for hundreds of metres. Instead, it’s a mess, with different varieties of ignimbrite jumbled together on the scale of tens of metres. For example, look at Figure 2, which is c.300 long. Having looked at outcrops that appeared in published papers with Jonathan Moles and ignimbrite expert Becky Williams we could not reconcile the complexities that were in front of our eyes with the simple logs in the published papers. This was very odd.

Several simple observations can be made from Figure 2: the base is not exposed; the top is exposed and is capped with sediments (which reassuringly contain clasts (i.e. chunks) of the Thórsmörk Ignimbrite – so the sediments are younger); there are pale and dark varieties; welding is variable; there is reworking of unwelded deposits. Yes, it really is a mess. The burning research question is – is this mess a mess related to original variable deposition of the PDCs, or is the mess due to later processes that moved separate ignimbrite domains in a way that brought them together?

So, let’s try and find some simpler exposures that might help us shed light on this variability. This is a common survival tactic when faced with a difficult field problem – go and look for simpler examples to understand first, then come back to the complex stuff.

Figure 3 (courtesy of Jonathan Moles) shows some of his mapping of southern Tindfjallajökull, where the Thórsmörk Ignimbrite occurs within a series of sedimentary units (tills) which – due to matrix and clast characteristics – we interpret as being glacial in origin. These units thin towards the margins of an ancient basin located in SE Tindfjallajökull, and in Figure 3 they can be seen thinning up to the right (towards the basin margin) with the ignimbrite sandwiched between tills. But is the complete ignimbrite preserved here?

No. Figure 4 shows that only one variety of ignimbrite is found – a part-welded dark phase. And although it looks simple from a distance, in detail it’s not as there are pods of other ignimbrite varieties around, such as the welded pale variety, and some unwelded ash. Yes, it’s still a puzzle.

But this area did demonstrate an interesting and previously-unknown feature of the local eruptive environment at the time the ignimbrite was being deposited – that a basin was filling with diamict (glacial sediments), and that this continued after the ignimbrite was deposited. The only difference between the diamicts below and above the ignimbrite is that the ones above contain clasts of ignimbrite.

A further look at ignimbrite exposures in this basin uncovered further complexity, including larger exposures of welded pale ignimbrite, further (rare) exposures of ash at the base, and that as well as being sandwiched within diamicts, the ignimbrite is also sitting directly on top of old basement rocks (pre-Tindfjallajökull), as shown in Figure 5. The latter merely indicates that the pre-ignimbrite topography was irregular, and that diamict was deposited irregularly.

Interesting insights into past environments that can be made from fieldwork in this area, is what happened after the ignimbrite was deposited. In many locations, subglacial basalts overlie the diamicts sitting on top of the ignimbrite (Figure 6). Logically, ignimbrite and diamict deposition would have required little or no ice to be present, but for subglacial basalts to be confined the ice must have been much thicker. So, it’s reasonable to conclude that a period of sustained cooling occurred after deposition of the ignimbrite. This points to a major environmental change after the eruption that formed the ignimbrite. See overlying subglacial basalts on Figure 6, and on Figures 3 and 4. And on Figure 7 (to come).

What else happened after the ignimbrite was erupted?

As well as the evidence for substantial subglacial basalt eruptions in SE Tindfjallajökull after diamicts had been deposited on top of the ignimbrite (as the basalts lie on top of the diamicts – see Figures 3, 4, 6, and 7), there is evidence in other locations that prior to subglacial basalt eruptions there was an episode of ‘reworking’ of the ignimbrite, involving especially the unwelded parts but also the welded part. In Figure 2 the label ‘reworked pale’ appears, and this reworked pale appears in many locations. Let’s look at a great exposure in Steinsholtsdalur, to the NE of Eyjafjallajökull.

In this location, there is the most spectacular and largest exposure of reworked Thórsmörk Ignimbrite that I have seen – but as I’ve not been everywhere, bigger and better ones may exist. At this exposure, reworked pale and unwelded pale ignimbrite occurs, and both are truncated by a subglacial basalt. The subglacial basalt is connected to small dykes that have intruded reworked ignimbrite, and in places the dykes have sintered (baked) the unwelded ignimbrite. Figure 7 shows Hugh Tuffen (@htuffen) and Lancaster University Masters student Alastair Hodgetts collecting sintered ignimbrite samples for lab analysis.

This is a marvellous exposure, and one worth spending days trying to decipher. For example, Figure 8 shows some detail of the reworked pale, with pumice-rich lenses and small faults affecting the finer-grained domains, and Figure 9 shows uppermost reworked pale cut by overlying subglacial basaltic volcaniclastics. There’s a great story to be uncovered.

Figure 10 reveals another puzzle about the reworked pale ignimbrite. Here, Becky Williams (@volcanologist) is near the base of the reworked pale, and a distinct shallow bedding can be seen. However, the bedding is dipping gently towards the east, which might imply deposition from a source to the west. Which is a puzzle, as there are no source volcanoes to the west that are known to erupt such compositions. Hekla’s magmas, for example, have quite different geochemical fingerprints.

I’ll take this opportunity to mention another puzzle: I’ve seen lots of reworked pale ignimbrite, but so far, I’ve not seen an exposure of reworked dark unwelded ignimbrite.

Moving on, there is ample evidence that the ignimbrite was eroded after being deposited. This is not surprising, given that we’ve seen evidence of thicker ice accumulating after ignimbrite emplacement (remember the overlying subglacial basalts). Like the ice currently forming Iceland’s ice caps and valley glaciers, this thicker ice would have been temperate (i.e. wet-based) ice, which is erosive. Much rarer polythermal (cold-based) ice, is much less erosive. A good example of erosion of a welded pale ignimbrite is seen in Figures 11 and 12, which are two views of the same exposure. This exposure also illustrates another notable feature that we see everywhere – the welded ignimbrite is always highly fractured, and not in a regular manner (e.g. columnar fractures). After welding, some unknown process has produced a series of irregular brittle fractures. This exposure permits the following time sequence to be suggested: ignimbrite emplacement, welding, erosion, then sediment deposition, then further erosion (as the sediments are themselves eroded). Interestingly, erosion of the welded ignimbrite occurred before the sediments were deposited, implying a time gap of unknown duration. And a final point – we don’t know what might be missing due to erosion, so the sequence above may be incomplete.

Ignimbrite Varieties

In this brief section, I’ll simply provide a few images of the different varieties of ignimbrite, and point out some of the puzzles to be solved. For me, the biggest and most crucial puzzle is the relationship between dark and pale ignimbrites – are they from the same eruption/source? Are they the same chemical composition, with the different colours related to a physical process (e.g. vesicle sizes/abundance)?

Figure 13 shows part-welded dark ignimbrite, whilst Figure 14 shows a foliated (or sheared) version of the same rock. There is considerable variation in lithics (non-juvenile clasts for the pedants) and in juvenile components (such as pumices), and Figure 15 (60 cm in length) shows a part-welded dark ignimbrite with abundant pumices and lithics, and note that some pale pumices contain lithics.

Figure 16 shows another oddity – a ‘spotted’ ignimbrite containing a mix of pale and dark domains. These are rare and only occur in small pods, and can be either welded or part-welded. Lab analysis would be needed to establish if there are two distinct (geochemical) components. Figure 17 shows a collection of welded to part-welded samples to illustrate just how variable this ignimbrite is.

Finally, we’ve found a few locations where pale and dark varieties are in close proximity. Figure 18 shows part-welded dark and pale ignimbrites in contact, with clearly different pumice contents (i.e. the dark is relatively poor in the white pumice clasts). This image is on top of a small cliff, and so the dark appears to be on top of the pale: if normal stratigraphic rules operate, the dark ignimbrite is younger. Interestingly, the layer immediately above the dark ignimbrite is rich in accretionary lapilli (Figure 19), and was the only good example we found. There are likely to be other exposures.

New evidence overturns an old model

I’m going to describe one bit of evidence that argues against Tindfjallajökull being the source of the Thórsmörk Ignimbrite. (There are other corroborating strands of evidence, but I’ll leave Jonathan Moles to tell the full story.)

The thickest exposure of the ignimbrite (c.200 m) logged in Karl Jørgensen’s 1980 classic paper was underneath a prominent rhyolite lava called Hestur on the lower SE flanks of Tindfjallajökull (Figure 20). This is almost 3 times thicker than any other measured thickness of the ignimbrite, so was I both curious and suspicious.

.

To access the pyroclastic pile requires crossing a fast-flowing knee-thigh deep glacial river (Figure 21), which swells in the afternoon as ice melting increases, which in turn increases adrenalin. The first exposures encountered near the stream lacked the essential components of the ignimbrite (e.g. the expected matrix, pumices, lithics, phenocrysts, etc), and I assumed that I’d encounter the true ignimbrite higher up. But no, all the way up to the lava base the lithology was the same – an impersistently bedded pumice-ash breccia containing occasional fresh obsidian clasts. Apart from a completely different componentry to the Thórsmörk Ignimbrite, this pyroclastic deposit was intruded by lava lobes displaying pervasive small columnar joints (Figure 22), something that I’ve never seen at any true Thórsmörk Ignimbrite locality, and is not reported in the literature. The contact between the pyroclastic pile and the overlying lava is seen in Figure 23, and is often oxidised, which usually indicates near-simultaneous eruption between such rocks (see Figure 20).

So, this is not the Thórsmörk Ignimbrite, and once you remove this abnormally high thickness from the pattern, the argument that Tindfjallajökull is the source weakens dramatically. But what is this pyroclastic deposit? I had the advantage of seeing similar pyroclastic deposits at the Mount Rainier and Öraefajökull volcanoes, where such deposits are found underneath rhyolite and dacite lavas that have flowed along ridges flanked by ice (valley glaciers). These pyroclastic deposits are thicker where there are sudden drops in the ridge, which allows relatively volatile-rich lava in the core of the lava to vesiculate. I’ve not yet published the Öraefajökull work, but the classic paper on ridge-bounded lavas in glacial settings at Mount Rainier is worth a read if you are interested – see Lescinsky & Sisson reference at the end.

Endnote

There’s much more to the Thórsmörk Ignimbrite than I’ve described here, and it remains something of an enigma – hence the title.

What you are getting in this article is an example of fieldwork ‘in the raw’ complete with uncertainties and puzzles. Published papers never reveal the process of reaching the final interpretations and conclusions as there is simply no space (always at a premium). Instead, in published papers you get a more polished version that is designed to be convincing.

You will note a relative lack of interpretation and speculation in this article. This is deliberate as I prefer much more corroborating evidence before suggesting interpretations in public, and we’re not at that stage yet. Of course, I have my ideas of what happened and how the different components are linked, but I’ll leave these in my notebooks for the moment.

So, I hope I’ve given you a bit of insight into a fascinating and accessible enigma. And one in which if you have the interest, you can go and have a look at yourself, as there are daily buses into Thórsmörk throughout the summer and a short walk will take you to some great exposures.

Supplementary Information – background

Prior to 2013, some of the key aspects of the Thórsmörk Ignimbrite within published peer-reviewed papers were:
1. The largest late Pleistocene ignimbrite in Iceland. At least 6 km3 in volume, which equates to 4 km3 DRE (dry rock equivalent = magma).
2. Up to 200 m thick, comprising multiple flow units.
3. Varies from unwelded to welded.
4. Irregularly exposed in valley bottoms in the Thórsmörk area, plus on the lower south-facing flanks of Tindfjallajökull volcano to the north.
5. Depositional environment involved little or no ice.
6. Geochemistry indicates that it’s a rhyolite. (For those who are interested, it’s a comendite, which is a mildly-peralkaline rhyolite.)
7. Tindfjallajökull was the source.
8. This eruption may have created the caldera at Tindfjallajökull.
9. Linked with North Atlantic Ash Zone II (NAAZII) which is a major marine marker horizon dated to c.53-58 ka which is roughly the middle of the last glacial period.
10. Meltout of ice-rafted debris was responsible. Therefore, there was sea ice to the south of Iceland.
11. One of the few ash layers visible in Greenland ice cores. Therefore, the eruption had to have an explosive and far-travelled atmospheric phase in addition to the terrain-hugging PDCs that formed the ignimbrite.
12. The ignimbrite itself has never been convincingly dated.

Subsequent work, largely by PhD student Jonathan Moles (The Open University) has shown:
• Point 2. Incorrect. Removing an incorrectly identified exposure reduces the maximum true thickness to c.70 m.
• Point 7. Tindfjallajökull is not the source.
• Point 8. The so-called caldera at Tindfjallajökull may not exist – no clear field evidence for downfaulting was found, despite good exposure.

Supplementary Information – useful references

Austin et al (2004). J Quaternary Sci, 19, 137-146.
Bramlette and Bradley (1941), USGS prof paper 196.
Jorgensen (1980). J Volcanol Geotherm Res, 8, 7-22.
Jorgensen (1987). Lithos, 20, 153-168.
Ram et al (1996). Geophys Res Letts, 23, 3167-3169
Sigurdsson et al (1998). EOS, 79, p.377.