Fakultät für Geowissenschaften - Digitale Hochschulschriften der LMU

Enhanced knowledge of pre- and syn-eruptive processes is vital to deal with the increasing threat imposed to population and infrastructure by volcanoes that have been active historically and may potentially erupt in future. For many years, most of this knowledge was received from experiments on analogue materials and/or numerical models. In order to increase their significance and applicability for the “real” case, the natural complexity may not be oversimplified and the input parameters must be reliable and realistic. In the light of this, a close connection of field and laboratory work is essential.

Volcanic eruptions may be phreatic, phreatomagmatic or magmatic, the latter scenario of which was addressed in this study. Rising magma is subject to decreasing lithostatic pressure. As a direct consequence, volatiles become increasingly oversaturated and bubbles will nucleate and grow depending on initial volatile content and melt viscosity. Both factors directly influence the diffusivity that limits the rate of bubble growth. Increasing amounts of bubbles increase the buoyancy difference to the surrounding rocks and lead to an acceleration of the rising melt-bubble mixture. Beside these limiting factors, the overpressure in the gas bubbles greatly depends on the magma’s ascent speed as it controls the residence time to conditions favourable to degassing (a combination of lithostatic pressure and magma temperature) and the time of overpressure reduction due to degassing. Volcanic eruptions occur when the bubbly melt can no longer withstand the exerted stress that derives from the overlying weight (lithostatic pressure), the expanding gas bubbles (internal gas overpressure) and different ascent velocities in the conduit (velocity profile). The melt will be fragmented and the gas-pyroclast mixture will be erupted.

This study has combined close investigation of the deposits of the 1990-1995 eruption of Unzen volcano, Japan and detailed laboratory investigations on samples of this eruption and other volcanoes. The field work intended to reveal the density distribution of samples from within the eruptive products. Although all samples already underwent one eruption, their physical state (e.g. crystallinity, porosity) mostly remained close to sub-surface pre-eruption conditions due to their high viscosity and accordingly allowed their usage for the analysis of the fragmentation behaviour. In that purpose, rapid decompression experiments that simulate volcanic eruptions triggered by internal gas overpressure have been performed at 850 °C to evaluate fragmentation threshold and fragmentation efficiency. Laboratory investigations of that kind are one approach to bridge the gap between observational field volcanology and risk assessment as they reveal information on what can not be investigated closely but what is essential to know for realistic eruption models and the adjacent hazard mitigation. Changing the experimental conditions and close investigation of the artificial products reveals the influence of physical properties on the fragmentation behaviour.

The density distribution inside a dome and the upper part of the conduit is crucial to the eruptive style of an explosive volcano. This information cannot be collected during an ongoing eruption but is important for future hazard assessment via modelling conduit flow and dome collapse/explosion behaviour. Therefore, the percentage of the mass fractions of all rock types in the primary and secondary volcanic deposits must be evaluated. For this purpose and at the lowest logistic effort, field-based density measurements have been performed on Unzen volcano, Japan. The resultant density distribution was found to be generally bimodal at constant peak values but changing peak ratios. The most abundant rock types at Unzen exhibited an open porosity of 8 and 20 vol.%, respectively. The porosity was found to be arranged in layers of cm- to dm-scale that were oriented subparallel to flow direc

Volcanic eruptions are one of the most spectacular and dangerous natural phenomena. Volcanic activity can either be effusive, dominated by quiescent emission of lava or explosive, dominated by the eruption of pyroclastic material. A rapid transition between these two regimes is possible. possible. On a broad scale, a clear distinction of the eruption style can be drawn by magma composition. Large-scale basaltic eruptions are mostly effusive, whereas large-scale silicic eruptions are mostly explosive. Thus silicic volcanism possesses a severe hazard potential and can have devastating effects on human population.

This work comprises experimental investigations of the fragmentation behavior of porous magma as well as of the propagation of elastic waves within this material. The analyses were conducted on samples from Unzen Volcano, Japan as well as samples from Soufrière Hills Volcano, Montserrat, West Indies.

The elastic wave velocities of differently porous Unzen dacite samples were investigated with the use of a cubic multi-anvil press. The main results of this study show that porosity (density) and texture affect the elastic properties of samples at a given temperature. In particular, it can be stated: (1) Seismic velocities increase with pressure due to compaction and closing of microcracks. The Vp anisotropy decreases with pressure for the same reason. (2) Increasing the temperature also leads to higher elastic wave velocities and lower anisotropies. This must be highlighted as the inverse behavior is documented for the majority of rocks. The effect may be linked to reduction of pore volume and further closing of microcracks due to reduction of cooling tensions. At 600 °C mean Vp values of 4.31 - 5.64 km/s and mean Vs* values of 2.20 - 3. 32 km/s could be determined. (3) The velocity anisotropy can be linked to the texture of the samples: Those with a high anisotropy show a pronounced shape-preferred orientation of phenocrysts and microcrystals, sometimes in addition to layering within the groundmass of the sample. Since the crystals are typically aligned parallel to walls of volcanic conduits, the velocity normal to the walls is likely to be reduced.

The data allows better estimates of the properties of silicic volcanic rocks at shallow depths within volcanoes, e.g. at conduit walls. These estimates are vital for computation of conduit models as well as the modelling of volcano seismic data, and may lead to an improved analysis of precursor phenomena in volcanic areas.

The physical properties of magma within volcanic conduits and domes are crucial for modelling eruptions. This study comprises a detailed investigation of the fragmentation behavior (threshold and propagation speed) of differently porous sets of dacitic and andesitic samples derived from Unzen Volcano, Japan and Soufrière Hills Volcano, Montserrat, West Indies. The experiments were performed with a shock-tube based fragmentation apparatus and pertain to the brittle fragmentation process. The results show a strong influence of the open porosity and the initial pressure on the fragmentation behavior. The speed of fragmentation follows a logarithmic relationship with the pressure difference, the fragmentation threshold an inversely proportional power-law relationship with increasing porosity. In this study fragmentation speed values ranging from 15 - 150 m/s were observed for applied pressure differences of up to 40 MPa and open porosities from 2.5 - 67.1 %.

The expansion of the pressurized gas in the vesicles largely provides the energy driving the fragmentation process. The fragmentation speed results of all analyzed samples show a close relationship to the energy density (fragmentation energy standardized to volume). A logarithmical increase of the propagation speed was observed with the energy density as soon as the energy threshold of 2.0 x 0.5 J/m³ was exceeded. The fragmentation speed is independent from the origin and composition of the samples

Trace elements (elements that constitute less than 0.1 wt.% of the analyzed sample) are important tracers for a great variety of processes in many research areas, such as biochemistry, medicine, semi-conductor and nano-technology, environmental science and geosciences (e.g., Anita et al. [2002]; Becker et al. [2004]; Barbante et al. [2004]; Tibi and Heumann [2003]). Accordingly, much scientific effort and financial resources are raised to develop new high-performance analytical techniques and methods for trace element analysis. In geosciences, the capabilities of trace elements analytics have not been used to its full potential because of the complex matrix of the analyzed samples (rocks) and the time consuming procedure to obtain high-quality trace element data by isotope dilution. Accordingly, the aim of this study is to develop of a new, easy-to-use and fast ID-method for the simultaneous determination of many trace elements in geological materials. In addition, the application of this new technique to the analysis of ocean island basalts (OIB) revealed that evolution of geochemical mantle heterogeneities (HIMU, EM-1, EM-2) is far more complex than perviously thought.

In the first part of this thesis, a multi-element technique for the simultaneous determination of 12 trace element concentrations in geological materials by combined isotope dilution (ID) sector field inductively coupled plasma mass spectrometry (SF-ICP-MS) following simple sample digestion is presented. The concentrations of additional 14 other trace elements have been obtained using the ID determined elements as internal standards. This method combines the advantages of ID (high precision and accuracy) with those of SF-ICP-MS (multi-element capability, fast sample processing without element separation) and overcomes the most prevailing drawbacks of ICP-MS (matrix effects and drift in sensitivity). Trace element concentration data for the geological reference material BHVO-1 (n = 5) reproduce to within 1-3% RSD with an accuracy of 1-2% relative to respective literature data for ID values and 2-3% for all other values. To test the overall performance of the method the technique has been applied to the analysis of 17 well-characterized geological reference materials from the United States Geological Survey (USGS), the Geological Survey of Japan (GSJ) and the International Association of Geoanalysts (IAG). The sample set also includes the new USGS reference glasses BCR-2G, BHVO-2G, and BIR-1G, as well as the MPI-DING reference glasses KL2-G and ML3B-G and the National Institute of Standards and Technology (NIST) SRM 612. Most data agree within 3-4% with respective literature data. The concentration data of USGS reference glasses agree in most cases with respective data of the original rock powder within the combined standard uncertainty of the method (2-3%), except the U concentration of BIR-1G, which shows a three times higher concentration compared to BIR-1.

In the second part of this thesis, this new method is used to determine the trace element concentrations of basaltic samples form the ocean islands St. Helena, Gough and Tristan da Cunha. The results are used to test the validity of established models concerning the trace element systematics of mantle heterogeneities. Since the early 1990's, recycling of altered oceanic crust together with small amounts of 'pelagic' and 'terrigeneous' sediments has become somewhat of a paradigm for explaining the geochemical and isotopic systematics in global OIB. The vastly increased number of data in the literature, in addition to new high-precision trace element data on samples from St. Helena, Gough, and Tristan da Cunha presented here (altogether more than 300 analyses from basalts from 15 key islands), reveals that the trace element systematics in enriched mantle (EM)-type OIB are far more complex than previously thought. In contrast to EM basalts, HIMU (high μ; μ = 238U/204Pb) basalts have remarkably uniform t

The Antarctic Peninsula forms part of a magmatic arc at least since Jurassic times. Magmatic dykes are essential elements of such arcs and intrude along zones of instability. In contrast to other hypabyssal intrusions and the effusive products of arc activity, dykes do not only reflect the geochemical characteristics of their magma source but also the tectonic parameters at the time of their emplacement.

The South Shetland Islands form an archipelago located at the northern tip of the Antarctic Peninsula and belong to this arc.

Areas of up to 100,000 m2 have been mapped at several locations of these islands, mainly on King George and Livingston Island. A structural analysis of the dykes and

the host rocks was carried out, and about 250 dykes

were sampled for further studies.

As deduced from field relationships, on Livingston Island six different intrusive events could be distinguished, on King George Island up to seven. This subdivision into different intrusive events is also well reflected by the geochemical data.

Analysis of the structural data of the dykes and their host rocks shows, that the tectonic stress field was not only very similar throughout the archipelago, but that moreover only minor changes of this stress field occurred during the time of dyke emplacement. This holds for all investigated areas in the South Shetland Islands.

The geochemical data (ICP-MS) reveal, that most dykes of the South Shetland Islands belong to a calc-alkaline, arc-related suite, ranging from basalts to highly differentiated rhyolites. However, especially during early stages of intrusive activity in the respective areas, also tholeiites occur.

Isotopic data (Sr, Nd, Pb) prove a strong crustal component during initial stages of magmatic activity, especially on Hurd Peninsula (Livingston Island). This crustal component decreased with time, accompanied by an increase of sedimentary input into the subduction zone.

The high amount of crustal contamination during the initial stages was probably due to a still unstretched continental crust.

Besides the continental crust underlying the South Shetland Islands, partial melts from the subducted sediments, fluids derived from the subducting plate and a depleted, heterogeneous mantle wedge contributed to arc magma genesis.

According to Ar-Ar datings on plagioclase separates and K-Ar (WR) age determinations, dyke intrusion was restricted to the Paleocene and Eocene. The dykes started to intrude around the Cretaceous/Paleogene boundary at Livingston Island. Only around the Thanetian/Ypresian boundary, dyke intrusion commenced also further NE at Nelson and King George Island, culminating during the Lutetian at 47-45 Ma in all investigated areas. Dyke intrusion then ceased in the latter areas but still continued at Livingston Island until the Priabonian.

Combining the information given by the tectonic and geochemical datasets, the time interval covered by the dykes obviously marks a period of geodynamic stability. This includes a stable geometry of the subduction zone and the corresponding parameters (subduction direction and velocity) during that time, as well as stable magma sources. The contribution of the respective sources (sediments, slab, mantle, crust) varied, but the sources themselves remained the same.

Very primitive, olivine tholeiitic dykes sampled on Penguin Island as a by-product of this work yielded an unexpectedly high Ar-Ar age (Tortonian), thus questioning the onset of rifting in Bransfield Strait during the Pliocene, as believed so far.

Der von häufigen Befeuchtungs- und Trocknungszyklen hervorgerufene physikalische Prozess des Quellens und Schwindens wird gegenwärtig als Hauptschadensursache für die Verwitterung von tonmineralreichen Sandsteinen angesehen.

Hauptziel der Arbeit war deshalb die Messung des bei Befeuchtung entstehenden Druckes und der Vergleich der erhaltenen Quelldruck Werte mit den Festigkeitskennwerten der untersuchten Sandsteine.

Die Kapillarsperre stellt ein alternatives Oberflächenabdichtungssystem für Deponien dar. Sie besteht aus einem geneigten Schichtenverbund, bei dem die feinkörnige Kapillarschicht (KS) von der gröberkörnigen Kapillarbruchschicht (KBS) unterlagert wird. Unter ungesättigten Bedingungen führt der ausgeprägte Texturunterschied an der Schichtgrenze zu einer Behinderung der vertikalen Wasserbewegung. Erst bei Überschreiten der lateralen Dränkapazität der KS setzt ein Wasserübertritt in die KBS ein. Bei Rückgang der Belastung regeneriert sich das System.

Zur Untersuchung des Wassertransports in der KS wurden in einem Versuchstank zum einen Markierungsversuche mit einem Cocktail aus den Fluoreszenztracern Na-Naphthionat (Na), Uranin (U), Pyranin (P) und Sulforhodamin B (SRB) durchgeführt, wobei die Tracereingabe als momentane Eingabe in Form eines Dirac-Impulses erfolgte (Tm-Versuche). Zum anderen wurde eine Lösung aus dem Fluoreszenztracer U und NaBr als Salztracer mittels kontinuierlicher Eingabe zugeführt (Tk-Versuche). Die Beregnungsraten betrugen jeweils 20 l/d bis 80 l/d. Zur Abschätzung des Sorptions- und Transportverhaltens der Tracer erfolgten ergänzend Batch-Versuche (Schüttelversuche) sowie Säulenversuche unter ungesättigten Bedingungen mit Durchflussraten zwischen 2 l/d und 80 l/d. Die Tracereingabe bei den Säulenversuchen wurde analog zu den Tankversuchen ausgeführt (Sm- bzw. Sk-Versuche).

In den Batch-Versuchen zeigt U eine geringe Sorptivität mit Verteilungskoeffizienten Kd bis max. 0,08 cm³/g, während die Kd-Werte von P (bis 1,18 cm³/g) und SRB (0,83 cm³/g bis 2,92 cm³/g) auf eine ausgeprägte Sorptionsneigung dieser beiden Fluoreszenzfarbstoffe hindeuten. Für Na ist keine messbare Wechselwirkung mit dem KS-Material festzustellen, so dass dieser Tracer als Referenz bei der Beurteilung des Transportverhaltens der Stoffe verwendet wird.

Das in den Säulenversuchen mit momentaner Eingabe (Sm) ermittelte Transportverhalten der Fluoreszenztracer entspricht den Ergebnissen der Batch-Versuche. Die höchsten Retardationsfaktoren RD im Vergleich zu Na zeigt SRB mit 1,6 bis 2,3, gefolgt von P mit 1,2 bis 1,6 und U mit max. 1,15. Die Rückgewinnung der Tracer beträgt nahezu 100%, so dass eine irreversible Sorption ausgeschlossen werden kann. Die Auswertung der Sm-Versuche mit dem Programm FIELD (Version 5, MALOSZEWSKI 2000) ergibt eine Aufsplittung des Tracerdurchgangs in drei bis sechs Teilkurven. Ursache ist eine vermutlich durch die mangelhafte Funktion der eingesetzten Verteilerplatte hervorgerufene Ungleichverteilung des Wassergehalts in der Säule, die experimentell nachgewiesen werden konnte.

Die Säulenversuche mit kontinuierlicher Eingabe (Sk) zeigen als wichtigstes Ergebnis ein nahezu identisches Verhalten von U und NaBr. Daher wird NaBr als geeignet für den Einsatz in den nachfolgenden Tankversuchen angesehen.

Auch in den Tankversuchen mit momentaner Eingabe (Tm-Versuche) ist bei der Modellierung mit FIELD eine Aufsplittung des Tracerdurchgangs in bis zu zehn Teilkurven festzustellen. Ursache ist in diesem Fall die für eine Kapillarsperre typische Wassergehaltsverteilung in der KS, die durch eine Zunahme des Wassergehalts zur Schichtgrenze hin sowie hangabwärts charakterisiert ist. Das Transportverhalten der Fluoreszenzfarbstoffe ergibt ein ähnliches Bild wie in den Sm-Versuchen. U weist mit einem Retardationsfaktor von max. 1,12 eine geringe Sorptivität auf, die ebenso wie die Transportverzögerung von P mit RD-Werten zwischen 1,07 und 1,41 den Werten aus den Sm-Versuche entspricht. Lediglich SRB zeigt eine deutlich erhöhte Retardation, die RD-Werte liegen zwischen 2,67 und 4,38. Die Rückgewinnung beträgt für Na und U ca. 100%, dagegen erreichen P und SRB aufgrund des ausgeprägten Tailings der Durchgangskurven nur Werte von 98 % bzw. 96 %.

Bei den Tankversuchen mit kontinuierlicher Tracereingabe (Tk-Versuche) erlaubt der Einbau von Leitfähigkeitssonden, die Veränderung der elektrischen Leitfähigkeit nach der Zugabe des Salzt