Skip to main content
European Commission logo print header

Multi-disciplinary monitoring, modelling and forecasting of volcanic hazard (MULTIMO)

Deliverables

Under MULTIMO, we developed a new method for remote sensing of volcanic gas emissions, principally of SO2 gas. Measurements of SO2 fluxes from volcanoes provide a key diagnostic of volcanic activity and hence are crucial to volcanic hazard assessment. We were the first to establish an alternative technology for ultraviolet sensing of volcanic emissions, finding a replacement for an outdated technology (the COSPEC instrument) which has since the early 1970s served volcano observatories and atmospheric scientists interested in volcano emissions. There has been very rapid uptake by the volcanological community of the new techniques following our active dissemination of results via international conferences and workshops, and publication in international journals (geophysics, volcanology). We were instrumental in assisting the Montserrat Volcano Observatory to set up the world's first remote SO2 flux surveillance system on a volcano. Other observatories worldwide are now borrowing this approach, including the INGV in Italy for surveillance of the Southern Italian volcanoes. The key innovations of the approach are the miniaturisation of the spectrometers, the simplicity of optical parts, the ease for automation and data telemetry, and the application of real-time data processing algorithms. Above all, the methodology permits high time resolution and also high spatial resolution measurements of volcanic plumes. In terms of time resolution, there is now the possibility for routine correlation of geochemical datastreams with results from geodesy and seismology. The expected benefits include improved volcanic hazard assessment and greatly enhanced characterisation of volcanic sulfur emissions from volcanoes worldwide in both time and space.
Generically, volcanic process belongs to the class of far from equilibrium systems, characterized by a slow increase of mass/energy, at a variable rate, that is suddenly released. This release defines what is known as volcanic eruption that can be associated to a phase change. Previous to the eruption, volcanic processes undergo a self organizing process, giving rise to geometrical and dynamical structures that should manifest as observables. These observables and its time variations would manifest as precursory phenomena. A possible precursory activity, always present in volcanic processes, but not always preceding an eruption, is provided by volcanic tremors. It has previously been shown that the influence of the medium properties in the information provided by a seismogram, and as a consequence a routine interpretation of spectral peaks in terms of linear models, could be misleading. We have worked out this problem in two directions: - A method has been devised to compute the (scattering) medium response in terms of oscillators and - By constructing a phenomenological model able to capture the characteristics of observations in both frequency and phase space domains. As the appearance of precursors is a dynamic process, any possible candidate has to be continuously measured, or, in other words, it has to be monitored and analysed in Real Time or in Quasi Real Time. A procedure for Real-Time analysis for any kind of continuous sequential time series has been developed. The hardware system has been tested in live seismic servers and it is based on the implementation of diverse algorithms run using the ordinary cron facilities of Linux based systems. The diverse algorithms have been tested a posteriori using data recorded in several volcanoes. In addition, the system has revealed as a very important tool to monitor not only incidences induced by volcanic activity but also any incidence related to malfunctioning. As a complement of the real time evolution of any parameter, suitable to be studied in Real-Time, a continuous comparison of the amplitudes of every spectral component has been also done. The application of this methodology can characterize several aspects of the current status. On one hand, the spectral levels in absence of transitory signals are clearly determined and, on the other hand, the identification of ranges of frequencies and reflected activity are emphasized.
We have developed software for the analysis and forward modelling of low-frequency volcanic earthquakes. These low-frequency events linked with the eruptive behaviour of the volcano and understanding their source and behaviour is important in investigating the eruptive dynamics of the volcano. Resonance of the magma conduit is modelled using a 2-D finite difference code, demonstrating the behaviour of a seismic wave trapped at the interface of the magma conduit and the country rock. Additional code calculates the magma properties along the length of the conduit, which vary realistically with depth and with the effects of unloading. The behaviour of the seismic wave under these changing conditions can then be analysed. Application of a realistic topography to see the effects on seismic wave propogation has also been developed, whilst further work on the Q factor (1/attenuation) gives a greater insight into the effects that the magma has on the seismic waves. Data analysis software has been developed which identifies and classifies seismic events, and provide a set of seismic parameters for statistical analysis.
A collection of Open Source routines, involving Linux, Apache, Zope, Plone, MySql, Scilab among the others, that allow to browse and first analyse multiparametric data stored in a heterogeneous format. From the user point of view the main advantage is the possibility of browsing through datasets recorded on different volcanoes, with different instruments, with different sampling frequencies, stored in different formats, all via a consistent, user-friendly interface that transparently runs queries to the database, gets the data from the main storage units, generates the graphs and produces dynamically generated web pages to interact with the user. And when we say user, we refer to a full class of people that maybe interested to using the dynamic web interface. In fact, the Zope application server allows definition of the user privileges, both for what concerns data access, and procedures. In this way we can offer a similar interface to the routine operator that does the monitoring, and to the researcher and to the volcano observatory director, but they will be able to act on different levels of detail on the data, perhaps using different sets of routines. Handling a volcanic crisis and doing research on volcanic data are completely different tasks. However, they share common problems to be solved. One of these is the handling of huge amounts of etherogeneous data. At the current time, this handling is done in a completely different way in any university and in any volcano observatory. This implies that e.g. in the moment of a crisis a researcher arriving at a foreign country to help the local observatory staff (if any...) has to find its way among the "usual" problems of how the data is stored, where is it, how it is time-stamped and so on, stealing precious time to more high level and important tasks such as the ones more linked to answering the questions: "What is going on?" and "What is going to happen next?". Even if the foreign researcher decides to use his/her own software, there is always the issue of how to convert the existing data to the format his/her software can handle. Moreover, there is also the issue of licensing, i.e. paying the license fees to the owners of the software, which most of the time is proprietary and commercial. And the so-called "pirating" is not really a nice nor fair solution! The building of a full and consistent set of tools based on the open source framework is therefore a major result that comes out of MULTIMO. We really hope that this is just the beginning, i.e. that the MULTIMO developed routines will be the base of a continuously growing environment that maybe further specialized and expanded. We consider particularly important the realization of a full dynamical web environment that allows for the execution and monitoring of the results of many different approaches, including the stochastic one, for which the commercial approach is the standard, probably because this kind of time series analysis was mostly developed for processing economic time series. In this case, the choice of "going open source" is one with particularly important social implications, especially for third world countries.
- Bubble growth during decompression of magma: Bubble growth during decompression of supersaturated melt is controlled by two dimensionless numbers: the ratio of the diffusive and the viscous time scales over the decompression time scale. We explored the conditions for diffusion and viscosity controlled growth and provided analytical solutions for some specific cases. The model solutions, including the division to the growth regimes as function of the two parameters, provide a fast tool for estimation of the state of erupting magma in terms of gas overpressure, supersaturation and gas volume fraction. The model results are in agreement with the conditions of Plinian explosive eruption (e.g. - Mt St Helens, 18 May 1980), where high gas overpressure is expected. The conditions of effusion of lava domes with sudden onset of explosive activity are also in agreement with the model predictions, mostly in equilibrium degassing and partly in overpressure conditions. The results were summarized in a paper published in JVGR. - The bulk viscosity of bubbly magma and applications for volcanic earthquakes: We derived expressions for bulk viscosity of suspension of gas bubbles in incompressible Newtonian liquid that exsolves volatiles. When such suspension is subjected to decompression it expands. The dilatational motion and the driving pressure can be used to define the bulk viscosity of the magma. The resulting bulk viscosity is highly non-linear. At the beginning of the expansion process, when gas exsolution is fast, the expansion rate grows exponentially while the driving pressure slightly decreases, which means that bulk viscosity is formally negative. This negative value reflects the release of the energy stored in the supersaturated liquid and its transfer to mechanical work during exsolution. This is very important since suspensions with negative bulk viscosity may amplify acoustic waves that travel through them. The results were summarized in a paper published in J. of Fluid Mechanics. The results indicate that following decompression, the conduit releases a lot of energy and at the same time absorbs seismic energy. We developed analytical solutions for the effective bulk viscosity, bubble radius, and gas pressure following the imposing of a sinusoidal pressure wave on a bubbly conduit. A numerical code was also adapted to model the situation. Preliminary results were presented in AGU-EGU meeting in France and at the IUGG meeting in Japan. - Cyclic activity at Soufrière Hills, Monserrat: degassing induced pressurization and stick-slip extrusion: The growth of lava domes is often associated with cyclic variations of ground deformation, seismicity and mass flux of gas and magma. These are commonly attributed to cyclic variations of magma pressure and flow patterns of the magma. We examine a model that describes the degassing of supersaturated magma following plugging of the conduit. As magma volume is confined, diffusion of the excess volatiles leads to buildup of gas pressure within the bubbles and to elastic stresses in the deformed conduit walls. The time scale of this process is controlled by diffusion and is of the order of minutes to hours, similar to the measured period of tilt cycles, which are attributed to pressurization of domes. When the difference between magma pressure and ambient pressure exceeds the static friction between the plug and the host rock, the plug starts to extrude, bubbles expand and overpressure is relaxed. The rate of extrusion is controlled by the magmatic pressure and by the friction along shear zones between the plug and the host rock. We estimate friction using rate and state dependence model and assuming a steady state approximation. The calculated relaxation time of magma overpressure depends on the depth of effective degassing, velocity and friction parameters and is comparable to deflation periods in tilt measurements. When the magma overpressure drops to the dynamic strength of the slip surfaces, the plug sticks to the conduit walls and blocks the vent. - Fragmentation of bubbly magma in visco-elastic regime: Explosive eruptions occur when magma fragments in a conduit. The fragmentation process means that the magma does not react as a fluid, but as an elastic substance. We developed a visco-elastic, spheri-symmetrical bubble growth model. The main controlling parameter is the ratio between the Maxwell relaxation time and the stress loading time, known as Deborah number (De). For large Deborah numbers (De>1) the magma will react elastically, and for small ones (De<1) the magma will react as a viscous fluid. For De~1 a bubble growth model that accounts for both viscous flow and elastic strain of the melt around the bubble should be applied. A preliminary numerical growth model was developed and the conditions under which an ascending magma may enter the visco-elastic regime were explored.
In order to compare the locations of tilt and hybrid earthquakes that occurred simultaneously at Montserrat, - A model for deformations: A fully three-dimensional boundary element method for elastic medium is used to re-interpret the tilt cycles measured in 1997 at Montserrat. Assuming the source of tilts is located along a cylindrical conduit, we show that this source is located 900 meters depth below the dome summit. Scaling the conduit diameter so that the elastic rock matrix does not fail, the measured tilt amplitudes are explained by a conduit diameter larger than 100 meters. This size is larger than the 30 meter diameter estimated from petrological and geological observations, but it can be explained if the medium around the conduit is visco-plastic. - A method to localize LPs at Montserrat was developed: The weak impulsive onset of hybrid earthquakes poses a challenge to conventional arrival time location algorithms. Considering the waveform similarity of events within earthquake swarms on Montserrat, their location are constrained by picks obtained from stacked seismograms. Using a velocity structure derived from fitting the extensive catalogue of events from the ongoing eruption, the events associated with the deformations cycles are located at approximately 1250 meters below the dome summit.
The aim of the project was to build a broad experimental database on the eruptive behaviour of volcanoes. The concept included an integrative field - laboratory - numerical model feedback approach. All segments were realised and due to the feedback experiments were realised following the results of the models and vice versa. Our results will have important implications for the mechanism of initiation and cessation of volcanic eruptions. The experimental results include magma properties as rheology a parameter necessary for models on flow and prediction of lava flows. With the characterisation of the fragmentation behaviour the strength of porous magma its fragmentation efficiency as well as the speed of the fragmentation process were analysed. These unique experimental data allowed a review on existing models. Our threshold curve can be used to predict how much overpressure is required to start explosive fragmentation of magma of known porosity. Current techniques for the geophysical and geochemical monitoring of active volcanoes provide reliable estimates of the pressurization state and depth of magma. This information serve as an input for numerical models able to calculate the porosity and pressure of the magma. In a scenario of expected dome collapse, sector collapse, or even in dyke opening events or vulcanian blasts. This approach, in combination with field density measurement of large pyroclasts, may help to better understand the local fragmentation dynamics of heterogeneous domes and conduit fillings. Besides the rocks pore volume and the gas pressure differential, the efficiency of degassing, i.e. the rock’s permeability, plays an important role considering the fragmentation behaviour of volcanic rocks. If the permeability is high enough, no overpressure develops; whereas at low magma permeability, gas flow is hindered and overpressure develops that may eventually lead to bubble wall failure and fragmentation. The degassing may occur through cracks or a network of interconnected bubbles. We realised during the experiments, that the permeability (k) is just a characterisation as it is measured in a laminar flow regime. The process of degassing during an eruption or violent degassing event is a forced turbulent flow that behaves different from the numerically used systems. Experimental data on the density controlled fragmentation speed in combination to textural observations on Montserrat merged to the theory of conduit implosions were the importance of density variations throughout the conduit could be demonstrated. An urgent need for a real 3D-numerical model is revealed were through a similar feedback mechanism as in the present study the physical background is experimentally counterchecked.
The main products of the work at the department of Earth Sciences at Bristol University are numerical codes to describe the flow in volcanic conduits in extrusive and explosive eruptions. The mathematical models of this work are being described in various publications that have been published already or have been submitted or are in preparation. Two main models have been developed: - Models of transient explosive flows in conduits following a major pressure change that initiates explosive eruptions. The model considers in particular the role of molecular diffusion in governing the characteristics of the eruption. A major finding is that there is a spectrum of explosive eruptions for short-lived explosions to long-lasting and sustained explosive discharges. - Models of conduit flows in extrusive eruptions that consider the several non-linear feedbacks in these systems. A particular development has been to include a much more realistic parameterisation of crystal growth kinetics in the models. The work confirms previous modelling work that volcanic systems can have multiple regimes and can be extraordinarily sensitive to slight changes of conditions near transition points between the regimes. The codes that have been developed within MULTIMO will continue to be refined and will be open to the scientific community. A major direction of future research will be to integrate these models with models of magma chamber dynamics and elastic deformations of volcanic conduits. MULTIMO researchers at Bristol are participating in an international modelling group that is comparing the output of such codes on volcanic flows.
For the purposes of eruption forecasting and hazard mitigation, a volcanic crisis may be represented as a staged progression of states of unrest, each with its own timescale and likelihoods of transition to other states (or to climactic eruption). If the state conditions can be interpreted physically, e.g. in terms of advancing materials failure, this knowledge could be used directly to inform decisions on alert level setting. A multi-state Markov process provides one simple model for defining states and for estimating rates of switching between states. However, for eruptive processes, such states are not directly observable and must be inferred from latent markers, such as seismic activity, gas output, deformation rates, etc, some of which may be contradictory. Interpretations of uncertain data will be liable to error, so a model is needed which can simultaneously estimate both elements: the transition likelihoods of a hidden process, and the probabilities of state misclassification. We describe the concept and underlying principles of continuous time hidden Markov models and demonstrate them in a decision-support context with a preliminary working implementation using MULTIMO data. Where multi-parameter streams of raw, processed or conditioned data of different kinds are available, these can be input in near real-time to appropriate Hidden Multi-state Markov models, the outputs of each providing their own objective analyses of eruptive state in probabilistic terms. These separate, multiple indicators of state can then be input into a Bayesian Belief Network framework for weighing and combining them as different strands of evidence, together with other observations, data, interpretations, and expert judgment.
The classification of different kinds of seismic transients and the analysis of their distribution in time are important information to identify the sources involved in the generation of seismic signals and define their activation in relation with eruptive activity. We chose the Artificial Neural Networks (ANN) for the classification of seismic transients as they are automatic systems, which excel in pattern classification. We followed progressive steps for our applications, starting with the evaluation of the most suitable form of data codification. Then, we proceeded with the selection of the seismic transients to analyse, the preparation of data sets for training and test, the critical revision of the results, and the re-analysis with enlarged data sets. Our final applications of ANN have demonstrated that this automatic tool can achieve good results for the classification of seismic transients, overcoming the drawbacks of subjectivity of human operators. ANN also allow to analyse large data sets which could hardly be handle manually. We successfully tested ANN topologies on data recorded at Soufrière Hills Volcano by the Montserrat Volcano Observatory. ANN may be also of great help for other volcano observatories worldwide, as they can be implemented for analyses of past and on-line data.

Searching for OpenAIRE data...

There was an error trying to search data from OpenAIRE

No results available