This thesis attempts to develop our understanding of volcanic magma plumbing systems and the magmatic processes that operate within them, such as fractional crystallisation, crustal partial melting, assimilation, and magma mixing. I utilise petrology, rock and mineral geochemistry, and isotope systematics to seek to improve our ability to forecast the eruptive frequency and style of active volcanoes, an aspect often lacking in current volcano monitoring efforts. In particular, magma reservoir dynamics are investigated from a mineral scale at Katla volcano in Iceland, to a sub-mineral scale at Merapi, Kelud, and Toba volcanoes in Indonesia. The magma plumbing architecture of Katla volcano on Iceland is explored in the first part of this thesis. Crystalline components within tephra and volcanic rock preserve a record of the physical and chemical evolution of a magma, and are analysed through oxygen isotopic and thermobarometric techniques to temporally constrain changes in reservoir depth and decode the petrogenesis of the lavas. We find both prolonged upper crustal magma storage and shallow level assimilation to be occurring at Katla. The results generated from combining these analytical strands reveal the potential for unpredictable explosive volcanism at this lively Icelandic volcano. The second part of this thesis examines the magma plumbing systems of Merapi, Kelud and Toba volcanoes of the Sunda arc in Indonesia at higher temporal and petrological resolution than possible for Katla (e.g., due to the crystal poor character of the rocks). For this part of the thesis, minerals were analysed in-situ to take advantage of sub-crystal scale isotopic variations in order to investigate processes of shallow-level assimilation in the build-up to particular eruptions. We find that intra-crystal analyses reveal an otherwise hidden differentiation history at these volcanoes, and establish a better understanding as to how they may have rapidly achieved a critical explosive state. The outcomes of this thesis therefore deepen our knowledge of evolutionary trends in magma plumbing system dynamics, and highlight the importance of understanding the geochemical processes that can prime a volcano for eruption. Lastly, I emphasise the vital contribution petrology can make in current volcano monitoring efforts.

Surface manifestations of hydrothermal fluids such as fumaroles and hot springs provide valuable information about the level of activity of a volcano during quiescent period. Geochemical study of gas and spring waters is useful to elaborate geochemical model for magmatic-hydrothermal system. Furthermore, temporal geochemical monitoring of these fluids with time provides a better understanding in processes occurring inside the volcano and can be useful to detect any changes in the activity of the magmatic-hydrothermal system. This thesis investigates two hydrothermal systems at Kelud and Papandayan volcanoes that are located at Java Island in Indonesia. Kelud is considered as one of the most dangerous volcanoes of Java because of its frequent eruptions. After the last eruption that occurred in 1990, a new lake rapidly filled the crater of Kelud volcano. Water samples collected since 1993 are near neutral Na-K chloride fluids and are typical of aged hydrothermal system where the acidity has been completely neutralized by fluid-rock interaction and where the emission of acid magmatic gases has stopped. Two sudden increases in lake temperature in 1996 and 2001 were accompanied by rapid changes in lake water compositions and suggest the existence of two hydrothermal systems feeding the lake: a shallow hydrothermal system dominated by Ca-Mg sulfate waters and a deepest aquifer with neutral alkali chloride waters. From 2001 to 2005, measurements of CO2 emitted by the surface of the lake were performed by using the accumulation chamber method modified in order to work at the surface of a crater lake. Two statistical methods were used to process data: the graphical statistical and stochastic simulation methods. The results of graphical statistical approach showed that two different degassing processes are acting at the lake surface: one corresponding to CO2 fluxes resulting from rising bubbles and the other corresponding to equilibrium diffusion of dissolved CO2 at the water-air surface. Total CO2 emission rate estimated by stochastic simulation ranges from 105 t/day for 2001 to 32 t/day for 2005. Thermal energy released by the lake was also estimated by using an energy balance model with a new constraint using the CO2 flux. The thermal flux decreased from 200 MW (2001) to 100 MW (2002) and then remained stable. Correlation between the chemical data of waters, the fluxes of CO2 and energy show that a constant decrease in the level of activity of the volcano since 1993 occurred although the lake temperature has been stable since 2003. Since the last magmatic eruption that occurred in 1772, phreatic eruptions occur on Papandayan volcano with the last one in 2002. The volcanic material ejected during this eruption is essentially made of altered rocks from within the hydrothermal system. The interaction of acid waters with the host rocks corresponds to an advanced argilic alteration. The chemical compositions of waters from Papandayan volcano and Kelud lake waters are contrasting. Indeed, the spring waters sampled since 1994 are acid sulfate-chloride waters and acid sulfate waters. The chemical and isotopic analyses of gases and waters suggest a significant magmatic contribution in SO2, HCl and HF to the hydrothermal system. The chemical composition of waters sampled after the 2002 eruption have provided information about origin of this eruption. Decrease in chloride concentration and in delta 34S of dissolved sulfates showed that the magmatic contribution in these fluids are less important and that the waters are likely to be formed by the condensation of steam (H2O, H2S) rising from a boiling aquifer.

Surface manifestations of hydrothermal fluids such as fumaroles and hot springs provide valuable information about the level of activity of a volcano during quiescent period. Geochemical study of gas and spring waters is useful to elaborate geochemical model for magmatic-hydrothermal system. Furthermore, temporal geochemical monitoring of these fluids with time provides a better understanding in processes occurring inside the volcano and can be useful to detect any changes in the activity of the magmatic-hydrothermal system. This thesis investigates two hydrothermal systems at Kelud and Papandayan volcanoes that are located at Java Island in Indonesia. Kelud is considered as one of the most dangerous volcanoes of Java because of its frequent eruptions. After the last eruption that occurred in 1990, a new lake rapidly filled the crater of Kelud volcano. Water samples collected since 1993 are near neutral Na-K chloride fluids and are typical of aged hydrothermal system where the acidity has been completely neutralized by fluid-rock interaction and where the emission of acid magmatic gases has stopped. Two sudden increases in lake temperature in 1996 and 2001 were accompanied by rapid changes in lake water compositions and suggest the existence of two hydrothermal systems feeding the lake: a shallow hydrothermal system dominated by Ca-Mg sulfate waters and a deepest aquifer with neutral alkali chloride waters. From 2001 to 2005, measurements of CO2 emitted by the surface of the lake were performed by using the accumulation chamber method modified in order to work at the surface of a crater lake. Two statistical methods were used to process data: the graphical statistical and stochastic simulation methods. The results of graphical statistical approach showed that two different degassing processes are acting at the lake surface: one corresponding to CO2 fluxes resulting from rising bubbles and the other corresponding to equilibrium diffusion of dissolved CO2 at the water-air surface. Total CO2 emission rate estimated by stochastic simulation ranges from 105 t/day for 2001 to 32 t/day for 2005. Thermal energy released by the lake was also estimated by using an energy balance model with a new constraint using the CO2 flux. The thermal flux decreased from 200 MW (2001) to 100 MW (2002) and then remained stable. Correlation between the chemical data of waters, the fluxes of CO2 and energy show that a constant decrease in the level of activity of the volcano since 1993 occurred although the lake temperature has been stable since 2003. Since the last magmatic eruption that occurred in 1772, phreatic eruptions occur on Papandayan volcano with the last one in 2002. The volcanic material ejected during this eruption is essentially made of altered rocks from within the hydrothermal system. The interaction of acid waters with the host rocks corresponds to an advanced argilic alteration. The chemical compositions of waters from Papandayan volcano and Kelud lake waters are contrasting. Indeed, the spring waters sampled since 1994 are acid sulfate-chloride waters and acid sulfate waters. The chemical and isotopic analyses of gases and waters suggest a significant magmatic contribution in SO2, HCl and HF to the hydrothermal system. The chemical composition of waters sampled after the 2002 eruption have provided information about origin of this eruption. Decrease in chloride concentration and in delta 34S of dissolved sulfates showed that the magmatic contribution in these fluids are less important and that the waters are likely to be formed by the condensation of steam (H2O, H2S) rising from a boiling aquifer.<p><p> / Doctorat en sciences, Spécialisation géologie / info:eu-repo/semantics/nonPublished