ALBERT

Beschreibung: River incision into bedrock is an important process in the context of landscape evolution. Climate can affect the river incision process in several ways. Firstly, precipitation is the main source of river discharge that is required to initiate the motion of river sediment and expose the riverbed to erosion. However, the relationship between river incision and river discharge is often non-linear, in that river discharge has to exceed a certain erosional threshold to mobilize bedload sediment and expose the riverbed to erosion. Whether this erosion threshold plays a significant role, or not, depends on the flux and grain size distribution of river sediment. Besides its effect on discharge, the climate’s influence on bedrock weathering also affects river incision, by changing bedrock erodibility as well as the grain size distribution of river sediment. Finally, climate sets discharge variability, which determines how often and by how much river discharge exceeds the erosion threshold. Discharge variability is typically high in arid regions, whereas river discharge is less variable in humid regions. Although it is evident that climate has an effect on river incision, studies that have investigated the effect of climate on 10Be-derived erosion rates – which in a steady state landscape equal river incision rates – have often found ambiguous relationships. This is most likely because other non-climatic factors (e.g., tectonic uplift rates, lithology, biota) interplay to obscure potential climatic trends. In this PhD thesis, I investigate the role of climate on various aspects of the river incision process in the Coastal Cordillera of central Chile. I focus on regions that are underlain by similar granodioritic lithology, but are exposed to contrasting climate regimes (arid, semi-arid, mediterranean, and humid-temperate). Using this approach, I aimed to reduce the variations in non-climatic factors that may obscure the climatic effect on erosion and river incision processes. I used in situ cosmogenic-10Be in river sediment to quantify erosion and river incision rates. In situ cosmogenic 10Be is produced in quartz grains in the upper few meters of the earth’s surface by high energy cosmic rays. The 10Be concentration reflects the time that grains are exposed to cosmic rays, which is proportional to the residence time of grains in the surface layer (i.e., inversely to the erosion rate). As a result of this, cosmogenic-10Be is frequently measured in river sediment to constrain catchment average erosion rates, which in steady state landscapes should equal river incision rates. In the first study (Chapter 3), I investigate grain size-dependent 10Be concentrations in river sediment. In most studies, the sand fraction of river sediment is used to measure 10Be-derived erosion rates, however, in catchments where 10Be concentrations vary between different grain size classes, this may result in biased erosion rate estimates. I investigate the controls of precipitation, hillslope angle, lithology and abrasion on grain size-dependent 10Be concentrations in Chile and in other landscapes around the world. I sampled 7 different grain size classes in 4 catchments located in the above-described climate regions in the Chilean Coastal Cordillera. The results reveal that regional precipitation regime affects grain size-dependent 10Be concentrations through its effect on 1) the depth of erosion processes, and 2) the depth of biotic soil mixing, which produces a constant 10Be concentration over depth. To put this in a broader perspective, I compiled 10Be concentrations across different grain sizes sampled at the same sample location for 73 catchments around the world. Based on this global compilation, I conclude that grain size-dependent changes in 10Be concentrations have a high likelihood of occurring in catchments with thin soil layers, where deep-seated erosion processes (e.g., landslides) excavate coarse grains from greater depth, where 10Be concentrations are lower. Typically, such catchment characteristics are found in landscapes that feature steep topography (〉25°) and high mean annual precipitation rates (〉2000 mm yr-1). I additionally find that the modification of the grain size distribution by fluvial abrasion can result in grain size-dependent 10Be concentrations. This mainly occurs in catchments with easily erodible lithologies and long sediment travel distances (〉2300–7000 m, depending on lithology). I conclude that roughly 50% of the previously published 10Be-derived catchment average erosion rates potentially contain a grain size bias, because the catchments feature one or more of the catchment characteristics that can lead to grain size-dependent 10Be concentrations. In the second study (Chapter 4), I investigated how climatic forcing can affect discharge variability, by studying El Niño Southern Oscillation (ENSO)-induced hydrological extremes along a climate gradient in central Chile (~28-42°S). This study focusses on discharge time series of 183 river catchments, that are located in the high elevation Andes and the low elevation coastal region, and feature different hydrological regimes: snowmelt-dominated (nival) versus rainfall-dominated (pluvial). The river discharge data shows that the hydrological response to ENSO differs strongly along the climate gradient and shows clear contrasts between basins with the nival and pluvial discharge regimes. The semi-arid region experiences the strongest river discharge anomalies during both El Niño (increasing discharge) and La Niña events (decreasing discharge), whereas the hydrological anomalies are the smallest in the humid-temperate region. Furthermore, the magnitude and frequency of extreme discharge events increases in the semi-arid and mediterranean regions during the warm and wet El Niño phase, whereas discharges in the humid-temperate region are most sensitive to rainfall deficits during La Niña events revealed by a higher frequency of low flow conditions. Snow dynamics introduce large contrasts in the hydrological response between basins with the nival and pluvial discharge regimes. First of all, snowmelt dynamics induce a delayed discharge peak. Snowmelt-dominated basins, experience the largest El Niño-induced discharge peak during the snowmelt season in summer, whereas the ENSO-induced climatic anomalies are most extreme during winter and autumn. Moreover, the discharge variability is lower for snowmelt-dominated basins because snowmelt produces non-flashy river discharge over a longer hydrological response time. Finally, basins with the nival-type of discharge regime are not as strongly affected by droughts than pluvial type of basins during La Niña, because snowmelt-generated runoff provides a minimum river discharge level. The results of this study reveal that ENSO-induced climatic and hydrological anomalies contribute strongly to the high discharge variability that has been observed in the semi-arid region. Which implies that ENSO has an important effect on river incision processes in the semi-arid region. Finally, I discuss the implications of the results of this study for water resource management in Chile. In the next study (Chapter 5), I investigated long-term catchment average erosion rates in catchments along a climate gradient in Chile. I sampled ~10 catchments in three of the four climate regions (semi-arid, mediterranean, and humid-temperate). I specifically selected catchments that feature differences in normalized channel steepness (ksn) between the catchments, which is a topographic metric that reflects tectonic uplift rates in a steady state landscape. The 10Be-derived catchment average erosion rates revealed an increasing trend with ksn. Besides this, however, a secondary influence of climate was evident: the slope of ksn-erosion rate relationships was markedly steeper for the humid-temperate region compared to the semi-arid region. In other words, for a given normalized channel steepness index, the highest erosion rates were observed for the humid-temperate region and the lowest for the semi-arid region. I compiled and recalculated previously published 10Be-derived erosion rates of ~150 catchments in Chile to compare my results to the large-scale erosional dynamics in Chile. While my new results agreed well with published erosion rates from other catchments in the Coastal Cordillera, erosion rates in the Andes are higher, which I posit is due to higher precipitation rates and steeper topography. Where previous studies had difficulties with depicting a consistent climatic signal along the latitudinal gradient, new data analysis suggests, that the erosion rates in Chile reflect a combined tectonic and climatic signal, which agrees with my own data from the Coastal Cordillera. The large degree of scatter in the compiled dataset is likely induced by non-climatic factors (e.g., lithology or biota). I conclude that, in Chile, the erosional efficiency increases with increasing precipitation, which provides empirical evidence for the understanding that arid landscapes have to become steeper than humid landscapes to reach erosion rates that equal tectonic uplift rates in a steady state landscape. In the final study (Chapter 6), I tested whether erosion thresholds, which are set by the amount and grain size of river sediment, play a significant role in the river incision processes in catchments in the Chilean Coastal Cordillera. I applied the stochastic-threshold stream power model, calibrated with field data, and compared the best fit model results to the 10Be-derived erosion rates and median grain sizes that I measured for each river catchment. The results reveal that erosion thresholds do play a role in the relatively gently sloping catchments of the Chilean Coastal Cordillera. A sensitivity test revealed that the modelled erosion rates of the semi-arid region rapidly decrease under erosion thresholds that are set by a grain size of 〉 1 cm. River incision still seems to occur in the humid-temperate region for erosion thresholds that are set by considerable grain sizes, but rapidly decrease for grain sizes of 〉10 cm. I conclude that in gently sloping basins the sensitivity of river incision rates to erosion thresholds strongly depends on river discharge, because the channel steepness is too low to facilitate the mobilization of river sediment. The results suggest that river incision occurs rather infrequently in the semi-arid region, whereas it occurs more continuously in the humid-temperate region. I have planned to further test the sensitivity of the model to input parameters and investigate the magnitude and reoccurrence time of river incision processes in the different climate regions of the Chilean Coastal Cordillera to validate these findings, and to better understand river incision processes in gently sloping landscapes. This is the first study that tests the threshold-behaviour of the process of river incision in gently sloping basins in regions that are exposed to different climates. To conclude, in this PhD thesis I investigated the effect of climate on several aspects that are relevant in the process of river incision. This study contributes to the general understanding of the effect of climate on landscape evolution in gently sloping mountain ranges that cover roughly ~15% of the Earth's surface.

Beschreibung: Gashydrate, auch „brennendes Eis“ genannt, sind faszinierende, eisähnliche Feststoffe, die aus Wasser- und Gasmolekülen aufgebaut sind und weltweit an allen aktiven und passiven Kontinentalhängen und in Permafrostgebieten vorkommen. Doch ihr unauffälliges Erscheinungsbild täuscht: Die Einschlussverbindungen können beachtliche Mengen Methangas enthalten. Daher besteht einerseits die Hoffnung auf einen möglichen neuen Energieträger und andererseits die Sorge um eine nicht zu unterschätzende Quelle an klimaschädlichem Methangas. Gashydrate, hat die neueste Forschung gezeigt, bieten zudem in vielen Bereichen industrieller Anwendung eine durchaus vielversprechende Alternative zu konventionellen Verfahren. Das vorliegende Buch gibt eine Einführung in die physikalisch-chemischen Grundlagen der Hydratbildung und die Strukturen der Gashydratphasen. Basierend auf diesem grundlegenden Verständnis erklärt es die natürlichen Gashydratvorkommen und zeichnet mögliche Methoden des Abbaus und der Gewinnung von Methangas auf. Es beleuchtet Risiken, die von den Gashydratvorkommen in der Natur ausgehen könnten, und führt in die Möglichkeiten der Nutzung dieser Einschlussverbindungen in verschiedenen industriellen Anwendungsbereichen wie z.B. der Aufbereitung von Abwässern oder der Speicherung von Gasen ein. Zielgruppe dieser kompakten Einführung in die verschiedenen Aspekte der Gashydratforschung sind Studierende der Chemie und Geowissenschaften, Ingenieure, Techniker oder auch Wissenschaftler.

Beschreibung: Recent development in seismological instrumentation has made possible the deployment of temporary and permanent seismic networks with large numbers of seismometer stations. These networks register an extensive amount of seismic data that is then available to the scientific community for carrying out different types of seismological studies. Producing reliable earthquake catalogs, based on the implementation of efficient automatic event location procedures, which can handle such extensive amount of seismic data, is one of the critical steps prior to perform most studies. Precise and maximally complete event catalogs become essential in regions on the Earth where tectonic plate subduction takes place, such as the Chilean convergent margin. In subduction zones, seismicity analysis can help to shed light on the different tectonic processes involved in the generation of great tsunamigenic megathrust earthquakes, which are usually devastating for the inhabitants of these hazardous regions. Particularly, insights on the conditions characterizing the frictional behavior of a subduction interplate fault, can be investigated in detail from seismicity. This ultimately leads to a better assessment of the earthquake hazard potential in a subduction zone. In my thesis, I apply a multistage automatic earthquake detection and location workflow to generate a high-resolution earthquake catalog for the northern Chile region. This automatic workflow is applied to seismic data recorded by the Integrated Plate Boundary Observatory Chile (IPOC) permanent seismic network, as well as by a number of complementary temporary deployments, adding up to 〉100 seismic stations. The resulting event catalog contains ~19,000 foreshocks, aftershocks and background seismicity occurring in the time interval between one month preceding and nine months following the 1 April 2014 M8.1 Iquique earthquake. I analyse the seismicity features, in combination with modelled coseismic slip and modelled static stress changes, as well as geodetically-derived afterslip and interseismic locking models, in order to investigate the small-scale frictional heterogeneities on the plate interface and interpret the seismotectonic behavior of the overlying continental forearc in northern Chile. Results from this analysis point towards a primarily along-dip segmentation of the frictional behavior along the plate interface. In the downdip direction, an aseismic velocity-strengthening segment of the interface and a frictionally transitional region underlay, respectively, an aseismic frontal prism and a transitional zone in the outer continental wedge. Deeper on the plate interface, the seismogenic segment coincides with the highest coseismic slip underlying the inner continental wedge. The velocity-weakening segment connects downdip to a frictionally heterogeneous region where aftershock seismicity (interpreted as conditionally stable) and highest afterslip patches (velocity-strengthening) anticorrelate. A sparsity of events, which may be a generic feature along the Chilean subduction, separates this heterogenous segment from the deepest interplate events, where presumably frictionally transitional behaviour predominates. With the aim of improving the efficiency and simplifying the implementation of the earthquake catalog generation workflow, I additionally present DeepPhasePick, an automatic method entirely based on systematically optimized deep neural networks that carries out the first stages involved in this workflow: detection and picking of P and S phases originating from local earthquakes. DeepPhasePick makes use of a convolutional neural network architecture to perform the phase detection on three-component seismograms. It then uses two recurrent neural networks to conduct the phase picking on the vertical component for P phases and on the two-horizontal components for S phases. The phase time onsets and their corresponding onset uncertainties returned by the phase picking stage can feed a phase associator algorithm in the next step of the earthquake location workflow. DeepPhasePick architectures are optimized and trained using 〉30,000 manually-picked seismic records extracted from two sets of event waveforms occurring ~7 and ~17 years before the 2014 M8.1 Iquique earthquake, in a region of northern Chile that partially overlaps with the area covered by the automatically-derived ~19,000 earthquake catalog described above. The algorithm is then tested on different test sets: a set of manually-picked records independent from the training set, 〉1,000,000 automatically-picked records taken from the ~19,000 earthquake catalog, and a set of 〉200,000 automatically-picked records taken from another recently published earthquake catalog for the northern Chile region. Results from these tests show that DeepPhasePick is able to detect seismic phases with high accuracy, as well as to predict phase time onsets with a precision comparable to the more conventional phase picking methods, be they manual or automatic. Moreover, I demonstrate that DeepPhasePick’s detection and picking abilities perform effectively not only on the largely lower-seismic noise data recorded in northern Chile, but also generalize to higher-seismic noise data recorded from a different tectonic regime in an urban region of Albania.

Beschreibung: Durch die fortlaufende Weiterentwicklung seismologischer Sensoren ist es heute möglich, temporär oder dauerhaft installierte seismische Netzwerke bestehend aus vielen Einzelsensoren aufzubauen. Diese Netzwerke akkumulieren große Datenmengen, welche sodann der seismologischen Community zur Verfügung stehen, um mit ihnen verschiedenste Studien durchzuführen. Um solche Studien mithilfe dieser umfangreichen Datenmengen in Angriff nehmen zu können, bedarf es der Erstellung akkurater Bebenkataloge mittels effizienter automatisierter Erdbeben-Lokationsmethoden. Solche präzise lokalisierten und möglichst kompletten Bebenkataloge sind vor Allem in Regionen der Erde essentiell, wo die Subduktion einer tektonischen Platte auftritt, so zum Beispiel entlang des Chilenischen Kontinentalrands. Seismizitätsanalyse in Subduktionszonen kann eine wichtige Rolle bei der Erforschung der verschiedenen aktiven Prozesse spielen, die große Erdbeben entlang der Megathrust-Verwerfung hervorrufen. Solche Beben, die wiederum oft Tsunamis auslösen, haben oft katastrophale Auswirkungen für die Bewohner der betroffenen Gegenden. Eine detaillierte Untersuchung der örtlichen Seismizität kann insbesondere zu Erkenntnissen über die Reibungseigenschaften der Plattengrenze und den Bedingungen, von denen diese abhängen, führen. In meiner Dissertationsschrift stelle ich die Anwendung eines mehrstufigen, automatisierten Bebendetektions- und Lokationsablaufs vor, der zur Erstellung eines hochaufgelösten Bebenkatalogs für Nordchile benutzt wurde. Dieser automatische Workflow wurde mit Daten angewendet, die vom permanenten seismischen Netzwerk des IPOC-Konsortiums (IPOC = Integrated Plate Boundary Observatory Chile) sowie von einigen zusätzlichen temporär installierten Netzwerken aufgezeichnet wurden, so dass insgesamt mehr als 100 Stationen verwendet wurden. Der erstellte Bebenkatalog enthält etwa 19,000 Beben, darunter viele Vor- und Nachbeben des Iquique-Bebens vom 1. April 2014 (Momentenmagnitude 8.1) sowie Hintergrundaktivität, Alles in einem Zeitraum von einem Monat vor bis neun Monate nach dem Iquiquebeben. Ich werte die Seismizitätsmerkmale zusammen mit Modellierungsergebnissen des coseismischen Versatzes und der Spannungsänderungen durch das Beben aus, und beziehe auch geodätisch ermittelte Modelle von Afterslip und der interseismischen Kopplung zwischen den Platten mit in die Analyse ein. So kann ich die kleinskaligen Heterogenitäten in der Reibungsfestigkeit der Plattengrenze untersuchen und die seismotektonischen Charakteristika der darüber liegenden Region des kontinentalen Forearcs Nordchiles interpretieren. Die Ergebnisse dieser Forschungstätigkeit zeigen eine Segmentierung in den Reibungeigenschaften der Plattengrenze primär in Richtung des Einfallens der abtauchenden Platte. Mit ansteigender Tiefe entlang der Plattengrenze treten dort zunächst eine aseismische Region auf, deren Rheologie velocity-strengthening ist, gefolgt von einer Region die als transitional bezeichnet werden kann. Diese befinden sich unterhalb eines aseismischen Frontal Prisms sowie des äußeren kontinentalen Keils, der konditional stabile Reibungseigenschaften aufweist. In größerer Tiefe befindet sich die seismogene Zone der Plattengrenze, die die größten coseismischen Versätze aufweist und unterhalb des inneren kontinentalen Keils liegt. Auf dieses Segment, das velocity-weakening aufweist, folgt eine heterogene Region, in der Nachbebenaktivität und die höchsten Afterslip-Werte räumlich antikorreliert zu beobachten sind. Darunter liegt eine Region deutlich verminderter Bebenaktivität, die möglicherweise überall entlang des chilenischen Kontinentalrands auftritt, sowie schlussendlich die tiefsten Interplattenbeben, die wohl in einer Tiefenregion mit transitionalen Reibungseigen schaften aufreten. Zusätzlich stelle ich DeepPhasePick vor, eine neue automatisierte Methode zum Picken der Ersteinsätze von P- und S-Wellen lokaler Beben, die komplett auf systematisch optimierten neuronalen Netzwerken beruht. Das Ziel dieser Methode ist es, die Effizienz der automatisierten Bebenkatalogsproduktion zu verbessern sowie diese Prozedur zu vereinfachen. DeepPhasePick benutzt tiefe neuronale Netzwerkarchitektur, um ankommende Phasen in Drei-Komponenten- Seismogrammen zu detektieren. Daraufhin verwendet es zwei rekurrente neuronale Netze, um die P-Phase auf der Vertikalkomponente und die S-Phase auf beiden Horizontalkomponenten zu picken. Die Ersteinsatzzeiten sowie ihre Unsicherheiten, die vom Programm ausgegeben werden, können sodann in einen Phasen-Assoziierungs-Algorithmus eingespeist werden, der den nächsten Schritt im Workflow der automatisierten Bebenlokation darstellt. Die Architektur von DeepPhasePick wird optimiert und trainiert mithilfe von 〉30,000 handgepickter Phasenankünfte, die aus zwei Datensätzen aus Nordchile extrahiert wurden, die Zeiträume ca. 7 und ca. 17 Jahre vor dem Iquiquebeben umfassen. Örtlich überlappen die Regionen dieser Datensätze mit dem oben beschriebenen Bebenkatalog, der ca. 19,000 automatisch ermittelte Bebenlokationen umfasst. Der Algorithmus wird dann mit verschiedenen Datensätzen getestet: mit einem Satz handgepickter Datenspuren, der unabhängig vom Trainingsdatensatz ist, mit 〉1,000,000 automatisch gepickter Spuren aus dem oben beschriebenen, etwa 19,000 Beben umfassenden Katalog, sowie mit einem weiteren Satz von 〉200,000 automatisch gepickter Spuren aus einem anderen, kürzlich publizierten Bebenkatalog aus der Region Nordchile. Die Ergebnisse dieser Tests zeigen, dass DeepPhasePick seismische Phasen sehr akkurat detektiert sowie die Ersteinsätze mit einer Präzision bestimmt, die derer konventioneller Phasenpick-Methoden (seien sie automatisch oder manuell) gleichkommt. Darüber hinaus zeige ich, dass die Detektions- und Pickingfähigkeiten von DeepPhasePick nicht nur für die Daten aus Nordchile, die sich durch sehr niedriges Hintergrundrauschen auszeichnen, gut funktionieren. Auch Daten mit höherem Noise-Niveau und aus einem gänzlich anderen tektonischen Kontext, nämlich aus einer urbanen Region in Albanien, werden von DeepPhasePick präzise ausgewertet.

Beschreibung: The George Fisher deposit is located in the northern Australian Carpentaria province, which is host to several of the world’s largest Zn and Pb mineral deposits. The annual metal production from George Fisher is crucial in order to meet the global demand for Zn and Pb. The main ore minerals at George Fisher are sphalerite (ZnS) and galena (PbS), which occur in stratabound, and in discordant, massive sulphide ore bodies together with pyrite (FeS2) and pyrrhotite (FeS). These massive sulphide ore bodies are hosted within carbonaceous, pyritic, calcareous, dolomitic siltstones and mudstones of the Paleoproterozoic Urquhart Shale Formation (ca. 1654 Ma), which is also host to two other world class base metal deposits (Mount Isa and Hilton). All three deposits are highly deformed, the textural relationships are complex, and there is a lack of indicator minerals to constrain the metamorphic grade of the Urquhart Shale. As a result, there has been considerable debate over (1) the processes that led to the accumulation of such huge amounts of base metal sulphides in the deposits of the area, and (2) the alteration footprint these processes have produced beyond the massive sulphide zones. In this project, petrographic observations across several scales (drill core logging down to backscatter-electron microscopy) were combined with whole rock, and in situ, mineralogical, geochemical, and isotopic analyses of representative samples from (1) four drill holes that intersected the main ore bodies at George Fisher, and also from (2) a drill hole that intersected the barren Urquhart Shale Formation. These data were collected in order to constrain the background heterogeneity that is inherent to rocks of the Urquhart Shale Formation (e.g., from background diagenetic processes). Using this framework, the nature of some of the hydrothermal processes responsible for ore formation were constrained. Particular emphasis was put on constraining (1) the accumulation processes of reduced sulphur, (2) the ore forming processes, and (3) the mineralogical and geochemical alteration footprint of the George Fisher deposit. In brief, the data from this study suggest that the Urquhart Shale Formation in the location we studied has not undergone regional greenschist metamorphism. The mineralogy and the paleoredox proxies (S-isotope data, Mo concentrations, rare earth elements in carbonate minerals) from the barren Urquhart Shale sequence can, therefore, be interpreted in a sedimentary and diagenetic context; and those data concur with deposition of the Urquhart Shale Formation in a ferruginous, marine environment, which is consistent with the current understanding for the Paleoproterozoic oceans. Using this background composition as base line the petrographic, mineralogical, geochemical, and isotopic data from George Fisher suggest that (1) ore formation occurred in multiple events during diagenesis and later deformation, that (2) reduced sulphur in the deposit was likely derived via thermochemical sulphate reduction and the recycling of sulphur from pre-ore diagenetic pyrite, and that (3) fluid-rock interaction of hot (〉200-250 °C), saline (Cl-rich), metal-bearing hydrothermal fluids with the Urquhart Shale Formation led to the dolomitization and replacement of pre-ore carbonate and the precipitation of base metal sulphides. Besides ore formation at George Fisher, these hydrothermal processes have resulted in mineralogical and bulk geochemical changes that include (1) albite, chlorite, and calcite depletion, (2) dolomite, phyllosilicate, and sulphide formation, (3) Na and Sr depletion, and (4) Tl and Mn enrichment relative to the barren host rocks. Furthermore, the fluid-rock interaction has led to light rare earth element (LREE) depletion in hydrothermal and hydrothermally altered carbonate minerals relative to whole rock and pre-ore carbonate LREE compositions. Overall, this project has provided new constraints on background diagenetic and hydrothermal processes, and footprints, in the Urquhart Shale Formation at George Fisher. Moreover, the findings from this study can help to further refine exploration models for Zn-Pb deposits in one of the world’s most important base metal provinces.