Rapid destabilization of deep, superhydrous magma prior to the largest known Plinian eruption of Cerro Machin volcano, Colombia

Abstract

A detailed stratigraphy of the pyroclastic fall deposits associated with Cerro Machin volcano (CMV) is presented following a previously defined categorization of the different eruptive units: El Espartillal, P0, P1, El Guaico, P2 and El Anillo. For the largest Plinian eruption in the sequence (P1), two lithofacies were distinguished on the basis of sedimentary features, grain size and componentry analysis. Early stages of the eruption could have been associated with vulcanian-type phases characterized by conduit/plug clearing explosions, producing a monolithologic lithic-rich laminated basal layer. The climactic event is represented by a white to grey, clast-supported, reverse to normally graded pumice-rich lapilli layer formed by a sustained eruptive column that gradually waned towards the end of the eruption. Associated deposits were identified up to 40 km from the vent. Pumice clasts from the most explosive phase were sampled along thedeposit layer in order to characterize storage conditions and ascent rates for the magma erupted. Pumice samples were classified as medium-K, calc-alkaline dacites (63-67 wt.% SiO2). The mineral assemblage includes plagioclase+amphibole+biotite+quartz and olivine and orthopyroxene (Fo 89-92) as accessory phases with amphibole overgrowths. Geothermobarometry of unzoned amphiboles, cores of reversely zoned crystals and rims of normally zoned crystals indicate a temperature range from 825±17°C and 913±45°C, a pressure range from 270±75 MPa to 1000±320 MPa indicating crystallization depths of 8-29 km. Thermobarometry of minor populations of unzoned amphibole crystals, cores of normally zoned crystals and rims of reversely zoned amphiboles show the same crystallization pressures as the dominant amphibole populations, but higher temperatures between 850±17°C and 978±29°C. The presence of these small populations of high-temperature amphiboles suggests a minor recharge event that did not drastically change the average crystallization conditions of the main dacitic reservoir but that could have been the trigger mechanism for the explosive eruption. CMV dacites display several geochemical signatures of adakites (low Y < 14 ppm, high Sr >700ppm, high Al2O3 > 16 wt.%, low MgO < 3 wt.%) which suggest that CMV magmas are produced by fractional crystallization of primitive hydrous magmas at the Moho boundary. The primitive magmas could have been the result of the interaction between Ba-enriched fluids dehydrated from the subducted slab with mantle peridotite in the mantle wedge. Three lines of evidence support the presence of superhydrous magma (containing >~8 wt% H2O) beneath Cerro Machin: 1) water concentrations of 2–11 wt% measured in plagioclase- and quartz-hosted melt inclusions; 2) the presence of Fo89-92 olivine rimmed by high Mg# amphibole; and 3) measurements of ~100–167 ppm H2O in plagioclase phenocrysts. Assuming a partition coefficient of 0.002, measured plagioclase crystals crystallized in a melt with 5-8 wt% H2O. Water-diffusion modelling in plagioclase crystals indicates a minimum time of 1 day for the magma to ascend from shallow depth (<250 MPa, ~5 km) before the eruption. Rapid ascent times are also suggested by the absence of breakdown rims in amphibole crystals which indicate a magma ascent timescale of <4 days from 8 km to the surface (Rutherford & Hill, 1993). Results from this study indicate that the 3600 yr BP eruption was preceded by rapid mobilization and ascent of superhydrous dacitic magma from mid- to deep-crustal storage regions beneath Cerro Machin. This thesis comprises eight appendices (A-H), Appendix A contains fieldwork information including: map of locations, a description of sample labels and schematic stratigraphic columns. In Appendix B, descriptions of componentry analysis are presented as well as pictures of the different grains identified under binocular observations. Appendix C is conformed by ten petrographic forms in which petrographic observations are detailed. This is followed by Appendix D in which whole rock chemistry of five samples is listed. In Appendix E, water measurements in melt inclusions are presented. In Appendix F, measurements of water in feldspar are detailed while in Appendix G, glass, melt inclusions, mineral chemistry and pictures of each crystal are presented, it also contains plagioclase-melt and amphibole-melt equilibrium tests and geothermobarometry calculations. Finally, Appendix H contains the link with the repository of the Matlab code used for volatile diffusion modelling in feldspar.