Inhaltszusammenfassung

Understanding the generation and stagnation of melt inside the lithosphere or asthenosphere is crucial to understand eruption processes on Earth. Recent advances in developing new thermodynamic melting models allow us to model realistic rock compositions and to investigate their properties (density, melt fraction, chemistry and mineralogy) with pressure (P) and temperature (T). Incorporating such thermodynamic data into a thermomechanical code is required to better understand the temporal and...Understanding the generation and stagnation of melt inside the lithosphere or asthenosphere is crucial to understand eruption processes on Earth. Recent advances in developing new thermodynamic melting models allow us to model realistic rock compositions and to investigate their properties (density, melt fraction, chemistry and mineralogy) with pressure (P) and temperature (T). Incorporating such thermodynamic data into a thermomechanical code is required to better understand the temporal and spatial evolution of magmatic systems. In this thesis, we develop different approaches that couple geodynamic with thermodynamic modeling to investigate the compositional evolution of magmatic systems, focusing on intraplate magmatism triggered by a rising mantle plume and on arc-related crustal magmatism. The mantle plume consists of a heterogeneous source composed of pyrolitic and metasomatized mantle. Decompression and the occurrence of hydrous mantle rocks cause melting in the rising plume at different stages. The generated melt is extracted, as a critical melt fraction is exceeded, and is emplaced within or on top of the crust. Compositional trends of the extracted melts (consisting of up to twelve oxides) are used to describe melting related processes by focusing on the K2O/Na2O ratio as an indicator for the rock type that has been molten. The results are compared with magmatic rocks of the West Eifel volcanic field (Germany) to propose a possible scenario to explain the observation that features a trend of high-enriched to less-enriched rocks with time. The computation of mantle phase diagrams is executed with pMELTS, under consideration of several rock depletion degrees at different P-T conditions. To decrease the number of precomputed phase diagrams, a Self-organizing Map (SOM) is used to define P-T ranges of similar bulk rock compositions. The chemical and the mineralogical evolution of the crust during arc magmatism are studied by incorporating a semi-analytical fracture algorithm into a geodynamic code to simulate dike/sill formation. Sill intrusion triggers the formation of dikes which propagate in the direction of the maximum principal stress. Each melt extraction event is leaving behind a depleted source with locally changeable rock compositions (10 oxides are considered) that is enhanced in case magma mixing is involved. To track this change in rock composition on each individual marker, tens of thousands of phase diagrams are precomputed using Perple_X. As the longevity of magma reservoirs is decisive for magma differentiation's efficiency, we investigate the conditions that keep regions of accumulated sills in a partially crystallized state (e.g., lower rock cohesion or deeper sill injection zone). Produced rocks are compared with igneous rocks from different arcs and the relevance of the different rock types for crust genesis is evaluated. Furthermore, we present an autonomous forecast approach to extend the phase diagram database with possible rock composition requests from geodynamic simulations. For this, new bulk rock compositions are determined within boundaries that are constrained manually or through principal components. A reduction of the sampling space is considered by applying a clustering algorithm to the database entries. The developed petro-thermo-mechanical code can be applied to a wide range of geological settings to study magmatic processes that result in the formation of different igneous rock types.» weiterlesen» einklappen

Autoren

Klassifikation

DDC Sachgruppe:
Geowissenschaften