Exploring the first second of a Strombolian volcanic eruption
1 Introduction 
Different types of volcanoes exhibit different styles of
eruptions, but all of them are characterized by the rise and expansion of gas, initially dissolved in the melt, that
finally leads to the transport of magma onto the earth surface or into the earth atmosphere. Understanding the different
types of volcanic eruptions is
an essential part of a successful hazard mitigation strategy. So far nearly all mechanisms suggested to be
operating inside the volcanic plumbing system and during the onset of an eruption are based on indirect
measurement techniques (e.g. seismic observations, acoustic observations, gas measurements etc.) i.e.
those methods are not targeting the in situ processes directly. In fact, so far not many techniques for the
direct observation of processes in volcanic conduits are available. Admittedly, in situ observations are
difficult to carry out, especially if one wants to learn more about strombolian, vulcanian or even plinian
eruptions due to the violence of those eruptions. Through this work we demonstrate that Doppler radar
is a technology to overcome this difficulty allowing observation of so far unmonitored processes and
thereby significantly expanding our knowledge of volcanic eruptions.
1.1 Motivation
Over the course of the last decade we have been developing and improving Doppler radar observation
technology for in situ observation of volcanic eruptions.
When we first started carrying out those
measurements in 1996 this was a new technique. A little more than 10 years later a french group
and ourselves have established this technique as a valuable tool for the observation of in situ processes
during volcanic eruptions and the data interpretation has also found applications in the interpretation
of weather radar observations of volcanic eruptions. One of the things I was most intrigued by this
method early on is that it would potentially allow a complete energy balance for strombolian eruptions,
something which was not possible by the beginning of this century. It took several more years to improve
instrumentation and data processing techniques as well as a lot of model development until finally this
goal was reached through this project.
1.2 Main goals of this study
The main questions we have tried to target with this proposal are:
• How do bubbles burst during a strombolian eruption?
• What is the overpressure in a bubble immediately prior to its burst?
• How much energy is involved in the burst of a bubble and how is it partitioned?
2 Results
In order to reach the goals defined above we carried out 2 six week long expeditions to Mt. Erebus,
Antarctica. For both expeditions we were invited by the scientists operating the Mt. Erebus Volcano
observatory. All cost for the expeditions starting in Christchurch, NZ, were covered by the National
Science Foundation, USA.
2.1 Field work
The first field campaign took place from 13 Nov 2005 until 6 Jan 2006, and the second one from
23.10.07 to 20.12.07. The first field campaign was very successful due to the high activity of Mt. Erebus.
The field crew consisted of several members from New Mexico Tech University (which operates the
Mount Erebus Volcano Observatory - MEVO), NASA Jet Propulsion Laboratory, Cambridge University,
and one member from Hamburg University (Alexander Gerst). All operations were based at the Lower Erebus Hut
(LEH), some 3 km away from the crater and summit region. The Volcano Doppler Radar (VDR) instruments and their accessories were
installed at three sites on the crater rim. The installation of three instead of only the two planned
VDRs, allowed the recording of eruptions from three different angles and with much higher sampling
rates than previously proposed.
The second field campaign was not as successful as the first one, mainly due to the very low activity
of the lava lake at Mt. Erebus during the field campaign in late 2007. During the observational period
only a few ma jor eruptions were observed. However, the expedition had one additional objective, which was
developed during 2007. The instrument in its current configuration records reflectivities for discrete
velocities, which are the displayed in so called velocity spectra. Usually
the reflectivity of non moving objects are of no interest as this is due to e.g. the echo of the crater wall
or in the case of Mt. Erebus due to the surface of the lava lake. Consequently this also is usually filtered
out. However, it does include some interesting and useful information, which we are now able to
extract.
In case the crater wall or the surface of the lava lake moves slowly the phase information of the reflected
wave can be used to detect very slow movements. The software of our fast radar has been updated
accordingly and we did record the phase information during the second campaign. However, processing
of the data is much more complicated and at this stage it is not 100% clear where the information, which
can be found in the data is coming from. Is it the deformation of the recording system (e.g. tripod,
antenna, radar mounting, etc.) or the a true signal from the lava lake. We are therefore working on an
improved calibration of the instrument and the accessories at the moment in order to able to filter out
the unwanted signal. We will keep on working on this issue (we have now three data sets which include
a deformation signal) and summarize these results in a publication at a later time.
2.2 Proof of concept of 3D measurements
In this project we developed a new timing system based
on an NTP (network time protocol) server. The whole system has been successfully tested before the
Erebus expedition during a quarry last. We were able to completely reconstruct the main blast direction.
Due to the geometry at Mt. Erebus a new processing technique had to be developed because the
observations at Mt. Erebus cannot be described as a directed jet as in case of the quarry blast or
explosions at Stromboli volcano. The new model is based on the bubble geometry observed at Mt. Erebus and also
allows us to determine the main direction of the bubble explosion.
The most important finding of this study with regard to the directivity of explosions is that the
rupture location of the bubble is random. There is no preferred location
like e.g. the apex of the bubble or the outer edge of the bubble. Admittedly the number of explosions,
which could be processed is small (due to the availability of the data) but we are confident that these
results would hold for even a larger number of explosions. This can actually be inferred from acoustic
observations for which much more data are available.
2.3 Explosion types
A detailed survey of all explosions recorded at Mt. Erebus volcano led to a classification of the different
events. The following two types of explosions could be identified:
• Type I explosions. They are characterized by the expansion of an initially intact membrane. The
acceleration of the membrane shows two characteristic peaks typically separated by 2-3 tenth of a
second.
• Type II explosions. In these explosions the membrane ruptures right at the start of the explosion
and the acceleration shows only one characteristic peak. Generally the velocities are higher than
for type I explosions.
There were two other explosion types (small explosions and so called blurred explosions) which were
not as common as the other ones.
The double peak accelerations which we observed in case of
the type I explosions are interpreted as two or more bubbles reaching the surface of the lava lake in a
short amount of time. It has been shown in laboratory experiments that when gas
slugs pass through a widening part of a conduit systems, a single slug can be separated into a couple of
smaller ones. We believe that this is the case for type I explosions.
For type II explosions the model developed is not really applicable as we do not observe an expansion
phase of the bubble prior to its burst. This is likely due to the fact that the expansion is so fast that it is
not captured by the radar in a couple of velocity spectrum. A detailed analysis of the type II explosions
is therefore not really possible.
2.4 The first second of a strombolian eruption
Our instrumentation was very fortunate for the project as this instrument allowed a temporal resolution of up to 25Hz.
The general process is an initial updoming of the
surface of the lava lake until the thinning shell above the bubble ruptures. Our radar captured
the velocity of the updoming magma shell. We further found a change in the velocity spectrum between the spectra.
One characteristic of the spectrum is that while the bubble is expanding there is a distinct cutoff velocity
due to the fastest moving part of the shell observed by the radar. After the burst of the bubble a large
number of fragments moved away from the rupture plane into different directions at very different
speeds leading to the observed broadening of the spectrum.
We also found that the expansion of the bubble is a continuous process and not an oscillation as has been
suggested by one suite of theoretical models for strombolian explosions. Our measurement alone clearly
showed that the model based on an oscillation prior to the bubble burst is wrong.
2.5 The energy balance of strombolian eruptions
Up to now no detailed energy balance regarding the burst of strombolian eruptions was possible because
there were no hard data on the actual bubble expansion prior to a strombolian eruption. This gap
was closed through our measurements with the Doppler radar. The energy balance included kinetic energy,
potential energy, dissipated energy, surface energy, acoustic
energy, seismic energy, and thermal energy. Most of the energy is actually kinetic energy, once the bubble
starts to expand. Only slightly before the bubble bursts potential and dissipated energy are getting
somewhere close to the kinetic energy being released but they are still one order of magnitude smaller
than the kinetic energy. Interestingly enough, acoustic and seismic energies are orders of magnitudes
smaller than the other forms of energy released.
2.6 Bubble overpressures
Through the energy balance we were also able to derive the bubble overpressures prior to their burst.
Average overpressures are found to be on the order of a few atmospheres. This shows that a moderate
overpressure is enough to account for the observed energy release described above. Furthermore this
estimate is within the wide range of previous estimates that can be found in the literature and it is
furthermore consistent with state of the art theoretical considerations and recent laboratory
experiments.
3 Outlook
This research marks a significant step forward in understanding the dynamics of Strombolian eruptions.
Through this project we were able to develop a consistent model for the pre-burst bubble growth
including an estimate of bubble pressures. Furthermore we can explain the waveform of the onset of
acoustic signals of strombolian eruptions and we have clearly shown that bubbles do not vibrate before
they burst but simply expand and burst.
Scientists
Matthias Hort
Alexander Gerst
Institut für Geophysik
Universität Hamburg
Research areas
Mt. Erebus, Antarctica
Publications
Gerst A, Hort M, Aster R, Johnson JB, 2009. The first second of a Strombolian volcanic eruption. Science to be submitted.
Gerst A, Hort M, Kyle PR, Vöge M, 2008. 4D velocity of Strombolian eruptions and man-made explosions derived from multiple Doppler
radar instruments. J. Volcanol. Geotherm. Res. 177, 648-660.
Homepage
---
Research funding organisation
German Research Foundation
Project number: HO 1411/16
Funding period: 2007 - 2010
