Hi G.G.J.Ernst,

Your review of previous work is a good one. I would point out a key distinction between pellet formation and the accretionary lapilli formation. The former often forms the core of the latter. The work of Richard Brown and Michael J. Branney has been critical in making this distinction because their accounts of the field associations revealed that pellets occur in both fall deposits and ignimbrites, whereas the accretionary lapilli (those with concentric rings of ash coating) only occur in ignimbrites formed from ground-hugging pyroclastic density currents. This distinction implies that the work of Gilbert, Lane, James and Durant are explorations of the pellet formation stage, which involves the adhering, cohesion or sticking of particles after fragmentation in the dilute plume environment. The accretionary layering is yet to be convincingly recreated.

Here at LMU (Munich) there's an ongoing research project (http://www.vertigo-itn.eu) in which Sebastian Mueller (http://www.vertigo-itn.eu/partners/researchers/mueller-sebastian) is undertaking an experimental study of ash aggregation. His work is combined with Glatt industries in Germany who provide industrial experience with granulation physics and surely is one way that accretionary lapilli can be formed from pellets (http://www.glatt.com/en/home). Sebastian has successfully formed pellets and is working on varying his flow parameters to achieve accretionary rings of ash around those pellets.

I encourage you to contact Sebastian Mueller.

Hope this is somewhat helpful :)

All the best,

Fabian

Hi Gerald,

As far as I know studies you have mentioned are the main ones on this kind of aggregates. There is also the group of Ulrich Kueppers in Germany, who used a fluidized bed for creating sub-spherical aggregates from volcanic ash and glass beads:

http://meetingorganizer.copernicus.org/EGU2015/EGU2015-3791-1.pdf

We are also working on volcanic aggregates with our wind tunnel, but at the moment we are focused on the so-call "dry aggregates" (i.e. particle clusters).

Cheers,

Mohsen

Dear Mohsen, dear Fabian,

Thank you very much for the insightful and informative answers.

I heard about the Geneva wind tunnel from Costanza (C. Bonadonna)  years ago and I am glad it is being used to study dry ash aggreation at least.

It seems to me that one kind of experimental lab study that would be useful would be to use one of the very few wind tunnels which have been used to study hailstone formation. Years ago, Bill Rose, Ray Shaw  and I tried to rescue one of those (around 2002 ?) for such a study but in the end we did not succeed in rescuing the setup. At the time, I have heard that there was one of those used by atmospheric scientists somewhere in Germany.

Fabian, thank you for pointing out Sebastien's work. I shall go and check this out.

As far as the ring structure is concerned, I carried out a very extensive SEM study (1500 SEM images) of dense  acc-lap structures (formed in dilute volcanic clouds, not pyroclastic density currents).

The conceptual model I have highlighted is fully consistent with this SEM and other data, both regarding the growth by riming (accretion of supercooled ash-filled drops around a frozen core) and regarding the crucial role of acc-lap recycling (fallout and reentrainment cycles). I am afraid this extensive research is unpublished. I presented the results during the IAVCEI Chile Conference in 2004, and this was well-received by everyone it seemed to me.

Previously to this, Adam Durant and I had done another very extensive study where we had measured up to 2000 acc-laps from a phreato-subplinian eruption and also studied their typical structure with SEM. Here already, it was clear that dense acc-laps form as volcanogenic hailstones. The ms was submitted to BV and was accepted (2002 ?) but the ideas we had proposed were rather new at the time and the referees asked for quite a bit of revisions (which were unjustified for a large part in my view). In the end, we got on with other work and the paper never came out.

One of the interesting aspect about this latter work is that it illustrates that wet accretion of ash-filled drops is limited by the drop break-up limit to a maximum size to about 5-6mm "diameter". Beyond this, the drops are torn apart by shear into smaller drops with a characteristic size of 1 (or 2) mm.

So if freezing does not intervene it does not seem possible to make dense ash aggregates exceeding 5-6 mm across.

Another key finding is that dense ash-balls are neither spherical nor strictly concentric. When one analyses the "concentricity", it is closely consistent with accretion and contact freezing of 1-2mm supercooled ash-filled drops.

When one analyses the aspect ratio of the ellipsoïdal ash balls (ie. very few are actually spheres), they very closely match the aspect ratios for hailstones.

Again this is consistent with dense accretionary lapilli (particularly all those in excess of 5-6mm across) forming as volcanogenic hailstones by riming beyond the initial wet accretion stage.

Analysis of field observations was also consistent with partial melting during fall for a large proportion of the frozen balls so that they were soft and deformed or splashed upon impact (forming what Mauro Rosi has elsewhere described as a versiculated tuff clearly related to overly wet accretionary lapilli). Other balls, typically the larger ones did not have time to melt and were still hard and solid so that some broke upon impact whereas others got embedded into the soft mud from molten acc-laps. 

Finally we also found that the frozen balls had a thin water film and that they fell through an ashcloud surge cloud filled with very fine ash and that the water coating selectively accreted these to form a thin outer layer that was distinctive as the outermost very fine ash  layer of the dense ash aggregates.

As to the distinction between pellets vs acc-laps, I do not remember if this correspond to those aggregates formed by wet accretion only versus those formed by wet accretion followed by a riming growth. I do not remember Rich and Mike making this distinction.

We also found other most interesting things, all consistent with the conceptual model I have highlighted,  but sadly the paper never came out...

I look forward to the first wind tunnel  experiments in the lab that may yet form all the features of dense acc-laps.

Thanks for the exchange.

Best,

Gerald

Hello all,

Yes, as far as I'm aware, dense, sub spherical accretionary lapilli have never been created in a wind tunnel outside of the Gilbert and Lane experiments. I suspect this is mainly because it's challenging to mobilize a moist, turbulent, high-concentration mixture of ash in a way that can be adequately characterized. But it certainly needs to be done. This is why we need to work with more engineers...

In my lab experiments using the vibratory pan method, which is basically a simple way to recreate a fluidized bed similar to the Schumacher and Schmincke approach, we were able to produce dense, subspherical aggregates and measure the sensitivity of their growth to different water contents. We  created multiple outer layers by moving the moist pellets into a new supply of fine ash (check out Fig. 9 e-f in the 2012 paper in Bull. Volcanol. linked below). It doesn't require a stretch of the imagination to consider the growth of multiple layers by recycling of aggregates into different regions of the airborne cloud system and PDCs. This has been clearly shown in field studies (Brown et al. 2010, Van Eaton and Wilson 2013) - aggregates interacting with PDCs have the most complex outer layers.

But I'm intrigued by what you're saying about ice, Gerald. This is something that motivated a lot of my PhD work, and I dedicated a section to the comparison of hailstones and accretionary lapilli (pp 19-21 in my thesis, linked below). In the end, I think there are definitely some parallels, in that the freezing of wet aggregates would temporarily stop growth (by reducing sticking efficiency), forming a sharp contact once the aggregate gets swept back into a warmer/wetter region of the plume or PDCs and more layers begin to grow.

However, I'm skeptical that riming in the form of dry growth is really important. Riming tends to form an openwork lattice in hailstones, whereas the outer layers of accretionary lapilli don't normally contain interstitial space. It looks like they form more like grain-by-grain accretion. For growth of the central cores (which are often vesicular, and poorly sorted), glazing may be a better analogue - more like a messy 'splat' of wet particles. In this case, most of the aggregate growth would occur in the wet regime (at least when we are talking about dense, sub spherical aggregates). Unlike raindrops, wet, poorly sorted ash has surprisingly high tensile strength, allowing them to stay stuck together even without precipitated salts or ice. I tried to bring this across in a paper last year called 'Hail formation triggers rapid ash aggregation in volcanic plumes'.

But regarding the ideas to experiment with ash aggregation in a cloud chamber, that would be a very cool project. I discussed this a little with Bill Rose when he visited CVO last year. It sounds like that setup is more or less up and running, right? In terms of getting the cloud chamber working at Michigan Tech? It would be great to have a little workshop there with all the people interested in aggregation, and brainstorm ways to use the cloud chamber. My initial questions would be if riming necessarily leaves behind a porous structure. And by how much the rate and extent of aggregation decreases in freezing conditions. It would also be interesting to measure the charge generation during ash-nucleated freezing, and development of a potential gradient in a turbulent, riming mixture. 

All the best,

Alexa

Article Growth of volcanic ash aggregates in the presence of liquid ...

Article The nature, origins and distribution of ash aggregates in a ...

Article Hail formation triggers rapid ash aggregation in volcanic plumes

Research Dynamics of large, wet volcanic clouds: The 25.4 ka Oruanui ...

Hello Alexa, Ulli, Volker and everyone,

Alexa, thank you for your suggestions and passing on the PDFs of your papers. As George Walker would have said, I believe, these are very fine papers because the work is anchored on a very large amount of fundamental systematic and systemic observations and measurements, ie. providing significant constraints for inverse modelling. Such studies are not so common so that it seems to me that they are particularly valuable and insightful. George, like Sherlock Holmes, never ignored any clue, and this is why such an approach can ultimately lead one to be left with only one possible model admissible with all the data collected from all possible angles.

I think that there is some confusion about acclaps in the sense that they can form in at least 3 different contexts and that we often see them in deposits the origin of which is not always 100% clear in publications.

1. One context is dense fluidized pyroclastic density currents. In such a context, rapid steam (or water if PDC is submarine one fluidized by seawater) streaming up through degassing channels can lead to the formation of some very large accretionary balls around a crystal or lithic (Ulli's experiments are also more pertinent to investigate those). It is not clear what may be the maximum size limit of these but one may readily imagine that it is degassing channel width. One may get insights into these by revisiting the milestone experiments by Colin (Colin Wilson) on fluidization of pyroclastic flow material and Colin or Steve Self or Richard Brown have seen/reported numerous examples of these in classic PDC deposits from New Zealand (Taupo, Oruanui...), the Philippines (eg. Taal) or PDC deposits from Tenerife. If I recall right, Bill Rose and friends have also seen good examples of those in PDC deposits from El Salvador (Ilopango ?).

A key point about these is that they do not fall through the air or experience recycling within eruption columns (eg. Ernst et al 1996, JGR) before leaving the umbrella base or eruption column side to then fall for say 10+ km to the ground. Of course, it seems likely that some airfall acclaps can also fall through PDC deposits (eg. Rich Brown work at Tenerife).

2. In the latter case (strictly airfall acclaps), there is just no way the acc-laps can grow to larger than the drop break-up limit, which for pure water is 6-8mm diameter (RA Houze, 1993, p80; Fig 3.6). Having an ash-filled drop does not affect the break-up limit very much based on a unpublished detailed study that Adam Durant and I conducted at Santorini on the US1 airfall deposits and which is quantitatively consistent with a wet accretion growth up to a break-up limit of 6mm (above such a limit size, probability of drop break-up soars exponentially rapidly so that 8mm ones formed purely by wet accretion are unlikely). So in the absence of freezing (and further growth by riming), airfall acclaps cannot exceed such a size (most of the acclaps observed by Costanza at SHV or those observed at Sakurajima in 1991 by Jenni Gilbert and Steve Lane typically do not exceeed this size of 6-7mm or so). Again, Ulli's expts at LMU can provide insights on growth and binding of acacclaps before freezing sets in.

If one produces SEM High Resolution maps of complete acclaps from such a context, all qualitative and quantitative parameters are consistent with further growth by accretion of ash-filled supercooled drops (with a characteristic size of 1-3mm or so; again this is constrained by the drop break-up limit).

I have made systematic and systemic observations of such acclaps and compared them with hailstones and there is little doubt that these acclaps form as volcanogenic hailstones. 

During hailstone fall, the stones undergo partial melting so that the largest balls may not melt due to rapid fall (10-30m/s say) but as they are brittle solids some will break upon impact (as in pumice breakage)...the majority of smaller balls melt and will produce mud rain but as partial melting is a stochastic process not all the balls will have melted, most will have partially melted and reverted to a partially frozen and partially ash-filled liquid drop like state so that upon impact all intermediates are observed between splashing (to make up the matrix of the acclap tuff bed), plastic deformation of a soft ball or brittle cracking or no breakage at all (soft landing upon mud-rain matrix made up of those acclaps that have melted before reaching the ground or would have splashed upon landing).

So the field association of all these types of features can indicate that one is dealing with airfall acclaps and not with PDC acclaps for example.  As these beds are dominantly mud-rain beds, slurry, slumping and secondary flood or lahar  erosion channels will also be seen, especially in near-vent/proximal deposits.

Further work on such airfall acclaps also leads to a quantitative relationship between maximum acclap size in a given acclap horizon and eruption column height which seems to be as helpful as the Carey and Sparks 1986 method to reconstruct Ht from deposit characteristics (unpublished data). This is great for phreatoplinian or phreatosubplinian past deposits to reconstruct Ht as these typically do not allow to apply the CS86 method and one only have thickness decay data to inverse model (eg. Bonadonna et al 1998).

For airfall  acclaps, the larger they are the higher column height; clearly this does not extend to the case of acclaps that form within degassing pipes in PDC deposits in a completely different way.

For airfall acclaps, the more severe the thunderstorm generated by the eruption (in this case, the eruption column is at the same time a severe thunderstorm), the more numerous the acclap horizons accumulate in rapid succession. In severe storms, numerous hailfall events occur whereas your small normal thunderstorm will only produce a single horizon of hailstones, perhaps 2 (this is also related to the lifetime of the thunderstorm which is short, but significantly  longer for a severe storm). So one expectation is that a sustained phreato(sub)plinian eruption may produce numerous acclap horizons in rapid succession whereas a weaker or short-lived event may produce a single acclap horizon per eruption burst event (eg. vulcanian).

3. A third context is that of eruptions producing maar-tuff-ring-tuff cone associations where large clast accumulating on a near vent ring avalanche into a slurry filling the vent (eg. see Pete Kokelaar; Jim Moore, Volker Lorenz). Multiple ejection-recycling -ejection-recycling into the vent (Surtseyan style eruptions) can form splendid armoured lapilli textures. Some great examples from Irazú 1963-65, Surtsey 1965 or so for example (classic papers by Guillermo Alvarado, and by Volker Lorenz for example). I am not saying that some of the other types were not also observed at Surtsey for example....they were: eruption plumes were also generated there....some acclaps were just like the US1 airfall acclaps (based on S Thorarinsson and by Jim Moore observations for example).

Finally, type 2, the partially melting frozen airfall acclap balls often fall not through clean air but through a lower air column filled with fine (most of it 20-40 microns) ash elutriated from base surges or denser PDC flows. Hence the last accreted ash layer on a falling acclap often  results from outer layer partial melting, shedding of most of that water due to shear during fast fall and size-selective wet accretion of very fine elutriated ash into an outer typically very thin water film (thickness of order of 1-2mm). These observations are for example consistent with the unpublished US1 study at Santorini.

Hoping this may help a little. and with comparing to your own observations and measurements relating to formation of dense ash aggregates (ie. acclaps).

Best wishes and kindest regards to all,

Gerald