Report of the Commission on Explosive Volcanism Short Course:
Recent Developments in Explosive Volcanism

23-24 August, 2008, Reykjavik, Iceland

by Rebecca Carey

Thanks to Sharon Allen, Shinji Takarada and Amanda Clarke for organizing this workshop and to the contributors listed below. There were approximately 80 participants for this 2-day short course, which was aimed to describe the current knowledge, developments and techniques in different fields of volcanology. The understanding of volcanic processes is rapidly increasing, and this workshop was an opportunity for volcanologists currently working in these fields to summarize and share these new developments with the community.

1. Bruce Houghton, University of Hawai'i – conduit dynamics, vesiculation and fragmentation processes

Bruce Houghton from the University of Hawai'i presented to the workshopees a clear step-by-step introduction to processes that occur in the conduit.

Bruce also emphasized that every process is linked to other processes and indicated to the group the complex picture of the feedback loops within the conduit.

In terms of future research Bruce indicated the following themes:

  1. It is critical for modeling purposes to understand the geometry and length of volcanic conduits to arrive at magma ascent rates. Currently conduit geometry is not imaged well, and that there is also the need to understand the mechanical properties of the conduits. Jeremy Phillips remarked that there is some understanding of the structure of conduits from kimberlite pipes, and that these studies might potentially become an important source.
  2. Conduit models commonly used range from 1D and 1.5D, and need to be further developed. Amanda Clarke pointed out that a talk given by Oleg Melnik, described 1-D and 2-D conduit models and indicated that 2-D models now need to be used for conduit modeling. However Amanda also remarked that there is still some understanding that can be achieved by making these assumptions in 1-D conduit flow models.

presentation file (PDF, 4.4MB)*
voice record file (mp3, 6.1MB)*

2. Ian skilling, University of Pittsburgh – Subglacial explosive volcanism

Ian Skilling from the University of Pittsburgh lectured the group on explosive subglacial eruptions. This field is in a development stage due to the two recent subglacial and emergent explosive eruptions that were well documented in Iceland (1996 Gjalp and 2004 Grimsvotn), and the accessibility of a wide array of older subglacial terrains in Iceland. Ian limited his presentation to describe only the types of eruptions that occur under an icecap. This type of setting severely affects the eruption style due to the pressure of the overlying ice and water. In addition to the pressure due to the volume of overlying ice, there are two more factors which control eruptive style and the landforms produced; the first is the volume of magma and the geometry of the vent system, the second is the ice surface and basement topography, which controls melt water drainage. The pressure in a subice cavity can also be less than glaciostatic and therefore have a very important influence on the behavior of the meltwater, overlying ice and nature of the products formed.

One problem for the subglacial explosive volcanism group of researchers is that their nomenclature is not clear, and often exact processes that occur in subaqueous and subglacial environments have different nomenclature. In addition, there are multiple words for similar landforms and processes. Similar to the submarine setting, there is complexity with understanding the pressure (critical thickness of overlying ice) at which magmatic volatile-driven explosive eruptions can take place.

Ian described the 1996 eruption of Gjalp underneath the Vatnajokull glacier, and the landforms and deposits that were produced during that eruption. In addition, he clarified some of the terms that volcanologists currently use when describing subglacial deposits use such as sediment gravity flows, high-density turbidites, density modified grain flows, slumps and slide scars, phreatomagmatic tephras. Lava pillow breccias, surtseyan tephras and tephra jet deposits.

In terms of identifying submarine settings from those of subglacial, Ian pointed out some key differences of the deposits such as, chaotic confined basin fills, and obvious ponds of volcanic which have been ponded against barriers (ice). From Ian's lecture it was very obvious that the deposits of suglacial eruptions are really complex, due to the ice and glacier structures which were once were present are are now missing. Ian illustrated these deposits with some great photographic examples of the deposits, and present-day subglacial settings in Iceland and Antarctica that are present-day analogues.

In terms of future research, there is a need to understand what the mechanisms of explosive eruptions are at depths where magmatic volatile-driven fragmentation should not occur (similar to the subaqueous setting). Need collaborations between glacial sedimentologists, glaciologists, structural geologists and volcanologists to understand how the subglacial landforms evolve.

presentation file (PDF, 7.1MB)*
voice record file (mp3, 7MB)*

3. Andy Woods, University of Cambridge – Eruption columns and plumes induced by volcanism

Andy Woods from the University of Cambridge presented a really thorough review of the physics of volcanic columns and the two-phase relationships in these plumes. He described the key processes that occur in different parts of the plume; the decompression driven jet, convective column and umbrella cloud. The timescales of processes such as entrainment, sedimentation and heat transfer, are time dependant, constantly changing and the effects (e.g., plume density, velocity) have feedback systems which are quite complex and intricate. The changes of these processes critically control the development of buoyant or collapsing columns.

The main points of his lecture covered the following themes;

  1. Understanding the conditions that control whether a plume will become buoyant or not.
  2. The implication of the plume processes, and feedback systems have for the deposits which are observed in the field (e.g. grading)
  3. The effects of the interaction of magma and water. The amount of water and thermal state of the magma are key factors which will determine whether the buoyancy of the plume is boosted, or subdued.
  4. The current development of 1-d and 3d plume models and what insights they hold for plume structure and the timescales of plume processes, e.g. entrainment.

1D analyses are invaluable for research quantifying simple first-order effects of variables in eruption plume dynamics, however are not sufficient to capture the complicated dynamics of an eruption column. Instead 2D and 3D computational simulations provide dimensional and temporal complexity, such as entrainment of air and mixing within different parts of an eruption column.

4. Jocelyn McPhie, University of Tasmania – Submarine explosive eruptions

Jocelyn McPhie from the University of Tasmania, Australia described the few recent developments in subaqueous Volcanology. The development of this field has been limited by the inaccessibility of the submarine realm, until recently where we now have video of some mafic submarine eruptions and also can collect in situ samples of submarine explosive eruption deposits. Jocelyn described the main styles of subaqueous eruptions; 1. Deep subaqueous magmatic volatile-driven; 2. Deep submarine steam-driven; 3. Shallow water phreatomagmatic explosive eruptions. Jocelyn also outlined the differences between the submarine and subaerial settings due to the physical properties of the water compared with air; water has significantly greater density (i.e., greater confining pressure on the system), greater heat capacity (i.e., cools the magma and the volcanic gas) and a greater viscosity (i.e., affects the behavior of particle motion in the ambient fluid). The additional complication is the fragmentation process, where magma: water interaction can contribute to fragmentation (steam explosions and quench fragmentation).

The depth limit of each submarine explosive style is difficult to constrain as it depends on the magma volatile content, magma rise rate and discharge rate. For the silicic end –1,500 m appears to be the limit for magmatic volatile driven fragmentation, but less clear for basalt, perhaps up to several thousands of meters of water depth.

The categories that Jocelyn described are listed below:

  1. Neptunian eruptions – a new term used to describe entirely submerged, plinian-like magmatic volatile-driven vesiculation and fragmentation with silicic magma compositions. The magmatic emergent steam cools and condenses leading to rapid decrease in the volume of the column and an increase in its density which inevitably causes column collapse. This effect is increased by the efficient mixing of the column with engulfing seawater. When erupted, pumice vesicles are filled with magmatic steam. During their cooling, condensation of the steam will trigger a suction effect, entraining the filling of the pumice interconnected vesicles by external water, giving a negative buoyancy to the clast. The filling of the pumice is proportional to clast size, the big pumices having a delayed water-logging. The grain size of the emergent mixture and the medium to which it is expelled leads to interesting sorting characteristics of the deposits. The ideal stratigraphic section through the deposits from this eruption style would be very thick density current deposits overlain by either thin beds of dilute density current deposits, simply ash or markedly bimodal deposits with giant pumice clasts at the top.

  2. Basaltic deep end-member-fountain, Strombolian-like and bulk interaction. Subaqueous fountains are driven primarily by magmatic volatiles as the primary driving force of fragmentation, however the water environment might also give rise to quench fragmentation and steam driven explosions (in shallow water). Strombolian-like: the nature of the juvenile clasts is the key to identification of this eruptive style on the ocean floor. This ejecta includes vesicular scoria with both blocky angular and fluidal morphologies, limu and glassy shards. The current idea is that these are explosions driven by volatile-rich supercritical magmatic fluids and not external water. The magmatic fluid is predominately CO2. The bulk interaction involves lava or magma engulfing water or wet sediment. The foundation of this idea is based on the lava entering the sea in Hawaii, producing a glassy thin plate of basaltic pyroclast fragment called limu, these pyroclasts are very thin plates of basaltic glass.

  3. Phreatomagmatic: Surtseyan and phreatoplinian for the basaltic and silicic end-members, respectively. These styles of eruption occur when vesiculating magma interacts with water in a shallow subaqueous setting. In this case, the column reaches the atmospheric surface and interacts with it.

presentation file (PDF, 6.2MB)*
voice record file (mp3, 8MB)*

5. Bruce Houghton, University of Hawai'i - Field techniques in explosive volcanism: quantitative approaches

This talk by Bruce Houghton from the University of Hawai'i described the techniques that volcanologists currently use to understand eruption dynamics. There are three scales: 10^2 to 10^4 m, such as deposit geometry and deposit thinning; mm to 1 m, such as bedding, grading, grain size and componentry; and micron to mm, such as bubble and crystal size distributions and melt compositions.

Bruce focused on the two larger scales: deposit geometry and the outcrop scale. For the deposit geometry, Bruce highlighted that the spacing between thickness measurements was critical to be able to successfully use empirical methods to calculate volume (e.g., Bonadonna and Houghton, 2005). The spacing suggested between measurement sites must not be greater than the thinning rate in that particular area. Isomass is a better technique then measuring thickness, but for really thick deposits, such as those from Plinian eruptions, this is often difficult especially in the proximal and medial areas.

He talked about the current methods that are used in the field and the two workshops that were designed to test the accuracy of our current methods. The first workshop was held by the tephra hazards working group, headed by Costanza Bonadonna and the second was held prior to the IAVCEI 2008 conference at Askja volcano. The results from both of these workshops highlighted that the current methods are very diverse between Volcanology groups, and hence the results are spectacularly variable. It highlights the need for a small paper or short note with a list and descriptions of the best methods to use for each technique (e.g. grain size, maximum pumice and lithic).

What is clear is that field volcanologists don't exactly know what and how to measure deposits in the field, which then provide the inputs into models. The data that field volcanologists supply to the modelers is currently not good enough, due to the variability of the techniques used. This can change with the planned short note in an international journal which clearly identifies field techniques. There are data that can't be retrieved if you don't do the full exercise, which can lead to expensive field trips with no reliable data!

On an outcrop scale, bedding and grading are two observations which have important inferences for eruption dynamics. Grading is a good way to look at shifts in eruption rate with time (as long as there are no changes in wind direction throughout the sedimentation period). By treating the deposit as a whole in the field (i.e. channel samples, or no observations of bedding or grading) the average eruption mass discharge is characterized, and this does not reflect the maximum intensity achieved or perhaps major changes in eruption dynamics. 1-d tephra sedimentation models for eruptions can ignore the presence or absence of grading, however time dependant models can start to consider these parameters, and these models will become very important for understanding conduit and plume dynamics with implications for tephra fall hazard mapping.

Bruce also explained clearly the preferred strategy for measuring clast density. It was described as follows; At least 100 clasts are required to characterize the heterogeneities of the density distribution and these clasts should be collected from between the 16 – 32 mm size fraction to avoid the complications of post fragmentation expansion. Bulk measurements of density and vesicularity are so fast and these can be used as a filter for other techniques you would like to use on the clasts, i.e., Microtextural studies.

presentation file (PDF, 3.9MB)*
voice record file (mp3, 8MB)*

6. Costanza Bonadonna, University of Geneva – Models of tephra dispersal

Costanza Bonadonna from the University of Geneva, Switzerland lectured about characterizing eruption deposits through field data acquisition techniques, and also discussed the current state and recent developments of numerical, analytical and empirical models used to characterize the dispersal of the eruption plume and sedimentation of particles.
Tephra deposits require characterization to derive information about eruptive parameters which ultimately link to hazard assessments. The key parameters are total eruption mass, eruption plume height, some form of measurement of eruption intensity (e.g., mass discharge rate), duration and total grain size distribution. The combination of field work, empirical, numerical and analytical modeling is now required.

In the past the methods to derive such parameters such as eruption plume height (e.g. Carey and Sparks 1986) have been used incorrectly. For example, the Carey and Sparks (1986) method to quantify eruption plume height is based on strong plume theory (strong plumes with Plinian-like plume velocities) and can't be used for example on weak bent over plumes or superbuoyant plumes. In addition, Carey and Sparks, (1986) clearly defined the method for measuring the lithic clasts in the field to arrive at the eruption plume height, however from the literature it is clear that many different methods have been implemented. The preliminary conclusions from the two CEV and Tephra Hazard group field workshops have also shown that depending on what clasts you measure and the number and way that you measure them, there can be very large discrepancies.

Costanza clarified what parameters need to be better described by more numerate and experimental folks as inputs to eruption plume models. The current models require better parameterization of processes that occur in the plume such as particle aggregation, in addition to particle shape and roughness and their effects on particle settling and plume dynamics. Field Volcanology folks also need to be familiar with the standard techniques (shortly to be standardized) in terms of field data acquisition, as these data are inputs into eruption plume models. If this does not occur then eruptions can't be compared to one another, plume models will not give good results, and numerical models cannot be used to validate other analytical and empirical models. Inversion techniques (2-d analytical models) have real promise in constraining eruptive parameters (e.g. total mass, column height). The combination of the inversion of deposit thickness and deposit grain size data can lead to good estimations of eruption plume height, mass discharge rate and duration. A combination of these techniques with the current empirical exponential thinning or power law models for eruptive volumes can lead to good calibration of the parameters, especially when the dataset is lacking data.

In terms of the current tephra dispersion models, which many are available in the literature, however each model has a particular focus and needs to be used for the right purpose. Tephra dispersion and sedimentation models solve the advection-diffusion equation (either with an analytical solution or numerically) for given eruption parameters at a given volcano. There are two different approaches in modeling tephra dispersion. The first approach consists in describing tephra sedimentation on the ground, while the second approach will track particle dispersion in the atmosphere.

1D analytical models: These models are developed to characterize tephra sedimentation from turbulent suspension and understand the depositional features along the dispersion axis in wind free conditions (e.g. Bursik et al., 1992; Bonadonna et al., 1998; Koyaguchi and Ohno, 2001) or with wind advection (Bonadonna and Phillips, 2003). These models will capture the rate of tephra thininng and break-in-slopes in tephra deposits.

2D and 3D analytical models: The output of these models (Hazmap, Ashfall, Tephra2) is in the form of isomass map of tephra accumulation on the ground. These models proved very useful in tephra fallout hazard assessments, as the output from this model can be in term of probability maps and and hazard curves.

3D (numerical, Fall3D): This model proposed by Costa et al. (2006) allows for the forecasting of atmospheric transport and deposition of volcanic ashes. Fall3D is a great improvement in tephra models, as it uses realistic 3D wind profiles to forecast the plume patterns and tephra sedimentation.

Particle tracking models (e.g. Puff will forecast the eruptive plume both in space and time after the beginning of the eruption (Searcy et al. 1998) and tephra sedimentation are also included in those models. But their main application is not in the forecasting tephra accumulation on the ground, but rather to forecast the cloud trajectory for aviation safety.

Future work: better characterization of tephra fallout – for example the total grain size distribution, particle shape, and particle aggregation need improve parameterization. The plume dynamics (mass distribution, plume velocity, plume geometry) has to be better described in analytical models, especially for more proximal deposits. We need standardization of the techniques in the field (e.g. measurement of the maximum clasts) and critical evaluation of the field data and observations. Model validation has to be carried out with field observations. Finally actual models of tephra dispersion don't incorporate proximal sedimentation either, and future models need to do so.

presentation file (PDF, 3MB)*


7. Mike Branney, University of Leicester – Pyroclastic density currents Part 1: interpreting deposits

Mike Branney from the University of Leicester, UK gave two talks describing pyroclastic density currents (pdc's). This first lecture by Mike outlined the complexity of studying and interpreting the deposits from pyroclastic density currents. The understanding of pyroclastic density current dynamics is not nearly as sophisticated as fall deposits, they are not able to be modeled entirely, however we do have a very good conceptual idea of the sorts of complications that may be occurring.

Pyroclastic density current deposits observed in the field do not represent the whole current, just a small portion of it (and likely the waning flow), and the deposits from a single flow may be absent all together. This has led to the over-interpretation of the flow behavior of pdc's. Mike introduced two new terms to the group; the first is depochron – which represent phases of deposition, and although likely invisible (unless there are compositional color changes) are helpful to understand how the flow progressed with time. The second term, entrachron is an entrainment isochron, a surface within a deposit that records clasts that were entrained in the current at the same moment in time.

Mike described a typical setting of a pyroclastic apron surrounding a volcano and the strategy for working on pdc deposits. Firstly those very good stratigraphic logs are required at many sites around the volcano because often the changes are subtle, and the nature of pdc's changes very rapidly and react very differently to topography. The pdc deposit can then be separated into eruption units, which are the first order divisions with hiatuses indicated by soils etc. in the succession. An eruption unit can then be subdivided into flow units, and fall layers might also be present within the stratigraphy. Mike cautioned that the subdivision into flow units is difficult and field workers must be cautious because what is being represented is only happening at this location. The flow unit can be studied in detail, and after many visited sites, flow units can be correlated to one another. One of the most important aspects is to separate the deposits into time frames, to work out the succession of events around the volcano. To understand the dynamics of these flows in the future, and to relate back to results stemming from models and analogue experiments, the best option is to study and map in excellent detail pdc deposits where the composition (and hence color) changes with time.

Experiments simulating pdc's have helped to understand the dynamics of these currents, but experiments still need to be further developed because they are not completely simulating the exact process. It is unlikely that models will be able to simulate all processes occurring in pdc's, however small scale experiments simulating individual flow processes, such as particle collisions can give vital information about how that particular factor may influences the moving current, which can then be fed into flow models.


8. Mike Branney, University of Leicester – Pyroclastic density currents Part II: transport and deposition

The second lecture of Mike's duo explored the support and transport mechanisms of pdc's. Volcanologists have had the tendency to believe that there is a single fluid support mechanism for the all of the particles observed within a pdc deposit that is studied in the field. However what is observed in the field does not represent the entire flow and there is a spectrum of many particle support mechanisms because the current is density stratified. There are two main factors why the transport mechanisms of pdc's are difficult to understand and model well: the first is that pdc's move along interfaces, so there is a certain amount of support due to the interface; the second is that pdc's are hyperconcentrated and tend to segregate themselves, which leads to difficulties in scaling and modeling.

There has always been the controversy over the terminology of PDC's and Mike's lecture today clarified and explained his current strategy towards this nomenclature. Firstly that all pyroclastic density currents simply be referred to as pdc's, this term simply relates to a pyroclastic density current which is a mixture of gases and pyroclastic particles. Following that there is a two-fold classification for all currents. Fully dilute and granular fluids. These terms are not the same as the terms pyroclastic surge, or pyroclastic flow. For example surges will have both granular and dilute facies. Unfortunately the deposits don't really hold any clues about whether granular or turbulent flow was dominant because you only see the basal part of the current deposited if any. Pdc's can also vary spatially and vertically adding to the complexity of their study and highlights that interpretations can't be pushed too far.

The things volcanologists currently have a reasonable understanding of are: granular fluid behavior, particle interactions and fluid transport, however the density stratification, the interface between granular and dilute, such highly polydispersed systems and the very high abundance of ash, make these flow difficult to understand their behavior. These processes need to be better parameterized through analogue experiments for inputs into future models. The range of processes and their overlapping time relationships most likely requires that individual processes need to be understood using analogue experiments, for example simulating the processes occurring in pdc's such as energy transfer during grain-grain interactions or heat transfer from particles to the interstitial gas etc. There has to be communication between field volcanologists, analogue experimentalists and modelers so that comparisons and interations exist between real deposits, modeling and simulations.

Mike stressed once more that one of the main problems is that there are no tracers to identify different pulses of currents to understand their geometry and how they flow over topography (with the exception of changes in composition). Thus we need to develop some detailed case studies of well-exposed young zoned pdc deposits to develop the special-temporal framework.


9. Shinji Takarada, Geological Survey of Japan – Debris Avalanches

Shinji Takarada from the Geological Survey of Japan gave the group an introduction to debris avalanches. He described the types of debris avalanches and the many different emplacement mechanisms (over 20 so far) that have been suggested for the different types of debris avalanches. Shinji presented some of the field areas that he had worked in such as Mount St Helens and presented the morphological characteristics of the source areas and the resulting deposits, in terms of volume, mobility and geometry.

The mechanisms that can trigger debris avalanches include for example, eruptive or intrusive activity, earthquake activity and major rainstorms. The Mount St Helens landslide greatly improved the understanding of debris avalanches as it was observed in real time and the source area is clear. In addition, debris avalanches in Montserrat have also been measured and observed. Although there are many debris avalanche deposits present in Japan (such as the 1792 Unzen debris avalanche), the actual processes involved in the collapse and the run out are generally only partly understood due to the lack of obvious source regions and observation data.

The features present in debris avalanche deposit terrains such as jigsaw cracks and hummocky surface morphologies are also presently not clearly explained. There are more than 20 models that have been proposed for debris avalanches, including lubrication, fluidization, grain flow, mud flow, viscous flow etc., and there is still much discussion and controversy within this field.

selected note (PDF, 2.7MB)*
presentation file (PDF, 15.7MB)*
voice record file (mp3, 7.5MB)*

10. Amanda Clarke, Arizona State University – Modeling explosive volcanic eruptions

Amanda focused upon current models of eruption plumes and depositional processes with a small introduction to conduit modeling within this lecture. After a brief introduction to the conservation equations which supplied the understanding of how small scale processes affect the equations, Amanda described the processes pre-fragmentation occurring in the conduit. These processes ultimately affect the emergent mixture and the processes thereafter.

Amanda described 1-D and multiphase models and the advantages of each to understand plume processes and what you can learn by using different models, i.e. understanding the interaction between different parameters and the first-order effects of variables. Amanda stressed that field data need to be collected for validation of the models and the simulations derived from these models.

Amanda also described multiphase 3D models, which include particles of different sizes and treat the gas and smaller particles as a continuum. These models yield considerable information by capturing the non-linear interactions and multidimensional behaviors of particles, such as the momentum exchange of particles related to differences in velocity, degree of thermal and mechanical coupling. These new multiphase models still need to be iteratively validated with experimental and field datasets.

Future research in the modeling fields that Amanda described are listed below:

Data collection from active eruptions – to understand the time-dependant processes. Some of the measurements of particular importance are for example, constraining the initial conditions, measuring sedimentation rates etc. These active measurements can be achieved with a greater development of the technology that can be applied, such as Doppler Radar.

Processes and properties that continue to be important to measure are; Entrainment, source conditions and unsteadiness, compressibility, particle interactions, particle aggregation and better understanding of the boundary layers.

presentation file (PDF, 2.6MB)*
voice record file (mp3, 7.4MB)*

11. Jeremy Phillips, University of Bristol – Analogue modeling

Jeremy Phillips from the University of Bristol talked about his research using analogue experiments to understand eruption and pyroclastic flow dynamics. Analogue experiments have been shown to be a real contribution to understanding the dynamics of such volcanic phenomena. Volcanology is becoming more process based and more people are interested in analogue experiments, there is a lot to learn from doing experiments!

The outline of Jeremy's lecture was a framework for how analogue experiments are done, how the data from the experiments is handled and how it related to what we see in the field. Analogue volcanic experiments are critical to understand volcanic processes and the key reason is what volcanologists see in the field is commonly what happens at the end of the process- the remnant. However, most of these processes are controlled by what happens at the source, which is what we don't know anything about, and can't get access to. With analogue experiments you can control things that are happening at the source, and change the rates of processes which is a really key way to learn about volcanic processes.

Jeremy explained that the analogue models use approximate relationships to describe the physics; analogue experiments use approximate physics by changing the starting materials. Because the experiments are being run at much smaller scales than reality, one of the issues that has to be overcome is scaling issues. Every parameter has to be scaled from the laboratory value to the full scale value.

Jeremy focused on two main topics; how experiments are scaled so the physics match and the experiment can be compared to the full scale process, and also dimensional analysis. He emphasized how to analyze data so that they have scale independence, so modelers can apply measurements, data and observations at the laboratory level to the field. He then finished with three examples outlining the natural process, experiments, data, extracting the scaling analysis and what it can tell you about the field.


*Selected notes, presentation and voice record files are personal use only. (All rights are reserved)

(Last update Dec. 21, 2008)  Top of the page