Refine
Has Fulltext
- yes (6048) (remove)
Year of publication
Document Type
- Postprint (2346)
- Doctoral Thesis (1787)
- Article (657)
- Preprint (425)
- Monograph/Edited Volume (249)
- Conference Proceeding (186)
- Working Paper (170)
- Master's Thesis (62)
- Habilitation Thesis (41)
- Review (36)
Language
- English (6048) (remove)
Keywords
- climate change (75)
- Klimawandel (52)
- machine learning (45)
- information structure (40)
- morphology (40)
- syntax (38)
- MOOC (37)
- e-learning (36)
- digital education (35)
- Curriculum Framework (34)
Institute
- Institut für Physik und Astronomie (651)
- Institut für Biochemie und Biologie (566)
- Extern (487)
- Mathematisch-Naturwissenschaftliche Fakultät (484)
- Institut für Geowissenschaften (477)
- Institut für Mathematik (477)
- Institut für Chemie (427)
- Department Linguistik (240)
- Hasso-Plattner-Institut für Digital Engineering GmbH (210)
- Institut für Umweltwissenschaften und Geographie (207)
This thesis uses attraction errors to study the potential usefulness of differential case in agreement processing. A common observation in speech production is that a noun other than the subject head may sometimes “usurp’’ the role of the agreement controller, leading to attraction errors such as *The key to the cabinets are rusty. A similar effect occurs in comprehension when the ungrammatical plural verb, preceded by the plural attractor, creates a false impression of grammaticality and is consequently read faster relative to conditions in which both key and cabinet are singular. One hypothesis is that attraction arises from the misretrieval of the attractor during the computation of grammatical relations between the subject head and the verb. If this holds true, then attraction should be less strong in contexts where case is a good pointer to which noun should control agreement on the verb.
The above hypothesis was tested in Modern Eastern Armenian, using relative clauses with non-intervening attractors. Eastern Armenian makes a useful distinction between definite animate subjects and non-subjects through discreet morphological marking (nominative and non-nominative, respectively), allowing for no syncretism. This property forms the basis of the experimental manipulations in this thesis.
A second aim of this thesis is to examine how comprehenders interpret the meaning of subject-verb agreement dependencies in contexts where the chances of interference are high. Given the interference from the attractor noun, do comprehenders misperceive the subject’s true number as plural (number misinterpretation), or do they misassign the thematic role associated with the subject position to the attractor (thematic misinterpretation)?
Whether attraction errors occur in Eastern Armenian and whether case information carried by the noun phrases is successfully used to match up the correct subject with the corresponding verb were examined through a forced-choice task and two self-paced reading experiments (Chapter 2).
Subsequently, adult speakers’ final interpretation of subject-verb agreement dependencies in contexts where an attractor is also present was examined using a self-paced reading task and follow-up open-ended questions that elicited free responses from participants (Chapter 3).
The results revealed clear attraction effects, but no consistent evidence that differential case modulated attraction. Attraction was not reduced in conditions where the subject and attractor took on different grammatical roles and bore differential case compared to when they both fulfilled a subject role and bore nominative case. Another finding was that the final interpretation of sentences with a distractor noun often resulted in errors. Singular subjects were frequently misperceived as plural in contexts where there was a plural attractor or a plural ungrammatical verb. Additionally, comprehenders often misperceived the attractor to be the thematic subject of the verb in conditions where the attractor was plural.
These findings are discussed within the context of retrieval-based accounts and encoding- based accounts that differ in their assumptions about how case may affect subject-verb agreement processing and their predictions about the meaning comprehenders derive from these dependencies.
The Beni Suef Basin, north-central Egypt, is one of several Mesozoic-Cenozoic rift basins of northeast Africa serving as the key for the successful petroleum exploration of Upper Egypt. It is bisected by the present-day Nile Valley into the East Beni Suef Basin (EBSB) and the West Beni Suef Basin (WBSB). Although the EBSB comprises the largest portion of the Beni Suef Basin, its petroleum system is not well understood. Thus, the lack of literature on the petroleum system of the EBSB compared with the WBSB motivates this study. This thesis aims to address the essential elements and processes of the Upper Cretaceous petroleum system of the EBSB through an integrated geological, geophysical, and geochemical data set and using the power of basin modeling techniques.
The source-generative potential of the sedimentary succession was evaluated using available TOC, Rock-Eval pyrolysis, and visual kerogen petrography analyses from three wells from the EBSB. Three kerogen types (II, II/III, and III) were recognized, where the Abu Roash “F” Member acts as the main source rock with good to excellent source potential, oil-prone mainly type II kerogen, and immature to marginal maturity levels. A 1D basin model was simulated at Tareef-1x Well from the EBSB to reconstruct the burial and thermal history of the basin. The burial history shows four alternating depositional and erosional phases from the Albian to the Miocene linked with the tectonic evolution of the basin. The Abu Roash “F” source rock in the EBSB is in the early oil window having reached a present-day transformation ratio of about 19 %. The magnitude of erosion and heat flow were tuned through a sensitivity analysis, where about 400 m of eroded thickness from the Apollonia Formation were estimated and the average heat flow was evaluated for the best prediction of thermal maturity. In addition, an improved 1D basin model at Fayoum-1X Well from the WBSB was developed, where two source rocks were identified; the Abu Roash “F” Member shows comparable characteristics to that of the EBSB, while the Lower Kharita Formation represents the main source rock in the WBSB.
In the second part of this thesis, an integrated geological and 2D basin modeling approach was employed with the main objective to investigate thermal maturity, hydrocarbon generation, migration, and accumulation at a basin scale. A geological conceptual model was developed by interpreting two intersecting seismic profiles tied with three exploration wells to define the geometrical architecture of the basin as a basis for the 2D basin models. Vitrinite reflectance measurements and corrected bottom hole temperature data were used to calibrate the developed 2D basin models. Based on the modeling results, the Abu Roash “F” source rock shows an increasing trend of temperature, thermal maturity, and transformation ratios from the basin margins to the center. Effective hydrocarbon generation occurred during the Upper Eocene following the trap formation in the Late Cretaceous. The kitchen area of hydrocarbon generation was defined at the deepest part of the basin showing the highest thermal maturity and hydrocarbon generation. Hydrocarbon migration occurred both laterally and vertically from the basin depocenter toward its margins, especially eastward to the structural traps, with the dominant vertical migration occurring primarily through faults and fracture systems that characterize petroleum systems of rift basins. In addition, another kitchen area at the northwest of the EBSB was located, which can explain the charging of Abu Roash “G” reservoirs. The models predict low hydrocarbon accumulations for the EBSB because of the ineffective sealing capacity of the overburden and leakage at the bounding normal faults.
The third part of the thesis tackled the structural framework and reservoir characterization of the Upper Cretaceous sedimentary succession of the EBSB using 29 seismic profiles and 5 boreholes. Eight key horizons were interpreted, and structural elements were delineated to construct the 3D framework model. The EBSB shows NW graben geometry that is dissected by a set of NW-SE and NNW-SSE normal fault system forming horsts and half grabens, which are attributed to the Early Cretaceous rifting tectonics linked with the opening of the Atlantic and Neotethys oceans. Compressional fold structures and strike-slip fault segments affected the Upper Cretaceous post-rift strata, which are related to the Late Cretaceous Syrian-Arc System tectonics.
Two clastic reservoirs were petrophysically identified in the Abu Roash “E” and “G” members. The average calculated petrophysical parameters include shale volume of about 7.4% and 10%, effective porosity of about 18% and 23%, water saturation of 34% and 42%, and net pay thickness of about 6.4 m and 3.8 m in the Abu Roash “E” and “G” reservoirs, respectively. The petrophysical variation is mainly controlled by facies variations of the two reservoirs, which in turn reflects their depositional environments. The Abu Roash “G” reservoir contains high amounts of argillaceous material and calcareous cement owing to its tidal flat depositional setting, hence showing poor reservoir quality compared with the Abu Roash “E” reservoir that exhibits clean sandstones of deltaic distributary mouth bars.
The presented results in this thesis can be applied to investigate the hydrocarbon potential of adjacent basins in the region (e.g., the Asyut and El Minya) which share a similar tectonostratigraphic evolutionary history with the Beni Suef Basin.
Stars are the most fundamental constituents of galaxies. The age, distribution, and composition of the stars in a galaxy can be used to derive the history and evolution of the host galaxy. The stars are responsible for synthesizing and distributing heavier elements. The understanding of the host stars is essential to decode the characteristics of planetary systems. This means that studying the formation, evolution, and death of stars is a central field of astrophysics.
Most stars are not born alone but are found to have one or several companions. In a significant fraction of those systems, the companions will interact with each other during their lifetime and so influence each other significantly. This can lead to substantial changes in the evolution of stars in close binary systems and cannot be neglected, if we want to understand stellar evolution. There are different kinds of interactions, which are observed. Tidal interactions can influence the stellar rotation velocity. Strong irradiation can alter the structure of the companions and is vital also for the understanding of close-in planets such as Hot Jupiters. Magnetic interactions can change the orbital period of the binary systems.
The most important interaction is mass transfer from one star to the other. Depending on the initial separation of the system, this can happen in different evolutionary phases. The mass ratio of both stars in the binary system is determining, if the mass transfer is stable or not. Binary systems with two stars of similar mass can experience stable mass transfer when they are close enough. Thereby mass is transferred from one star to the other, changing both stellar masses. The companion accreting the matter is spun-up and so rejuvenated. In systems where the stars have significantly different masses, the mass transfer is expected to happen on a dynamical timescale and the rate of mass transfer will be so high that a common envelope around both stars is formed. Friction in the envelope leads to a rapid shrinking of the orbit on the timescale of a few thousands of days and is difficult to observe. The observation of many evolved systems with orbital separations smaller than the radius of a red giant shows that such a phase must exist. During the spiral-in, energy and angular momentum is transferred to the envelope and can lead to its ejection, when the transferred energy is sufficient to unbind it.
Common-envelope evolution is crucial to understand many exciting systems, which are ob- served in our universe. It is essential for the formation of stellar-mass gravitational merger sources, as it can bring compact-object binaries close enough together so that gravitational waves can lead to a merger within a Hubble time. This phase is also vital for the understanding of progenitor systems of supernovae type Ia, which are used as standard candles to derive distances to far galaxies. With their help the existence of cosmological dark energy could be inferred. Supernovae type Ia are thought to be produced by the explosion of carbon-oxygen white dwarfs exceeding their upper mass limit, the Chandrasekhar mass. The most likely progenitor systems are still under debate.
The common envelope phase is one of the most significant and least-constrained processes in stellar binary evolution, and one of the most important unsolved problems in the understanding of stellar evolution. As it is a short-lived phase, direct observations are very unlikely and a statistically significant sample of systems after this phase is necessary to gain insight into it. For the physical understanding of this process (magneto-) hydrodynamical simulations have to be performed. A large enough sample of post-common envelope systems with observed masses, radii and orbital separations can provide the parameter range that has to be explained by the simulations.
In this thesis, I make a significant contribution to the investigation of binary systems after the common envelope phase by newly discovering a large sample post-common envelope systems and also significantly increasing the number of systems with derived parameters. This sample is a first step to acquire a statistically significant sample for a better understanding of this crucial but not yet understood phase. Moreover, I also present the detailed analysis of several kinds of interesting close binary systems studying different aspects of these systems, as the influence of tidal forces, the effect of mass transfer and orbital period changes among other things.
I will give a short introduction to the current state of knowledge regarding stellar evolution of single stars and the influence of close companions on this evolution. Moreover, I give a short introduction into the different kinds of close binary systems we studied. I also give a short summary of the methods used to investigate those systems. Subsequently, to the appended papers, I will discuss the results and give a short summary.
Multivariate time series are a form of real-valued sequence data that simultaneously record different time-dependent variables. Multivariate time series originate mostly from multi-sensor setups and serve a variety of important analytical purposes, including the detection of normal and abnormal behavior. Anomalies often occur in individual channels of a time series, but can also be found in the correlation of multiple channels. While effective data mining algorithms exist for the detection of anomalous and structurally conspicuous test recordings, these algorithms do not perform any semantic labelling. So, data analysts spend many hours connecting the large amounts of automatically extracted observations to their underlying root causes. The complexity, amount and variety of extracted time series make this task hard not only for humans, but also for existing algorithms: These algorithms either require training data for supervised learning, cannot deal with varying time series lengths, or suffer from exceptionally long runtimes.
To facilitate the analysis of anomalies in large and multivariate time series, we investigate three types of algorithms in this dissertation: Anomaly Detection, Clustering, and Classification. More precisely, we create an overview of the time series anomaly detection research field and point out short commings with published benchmarks. Then, we propose a novel and scalable time series anomaly detector that can find anomalies in the correlations of time series channels and reveal in which channels anomalies occur. To distribute the anomaly detection computation, we developed a novel library for building reactive and distributed algorithms. Moreover, we propose a fast and effective clustering technique for time series with varying lengths and introduce a framework for counteracting extremely skewed data partitions during the distributed training of machine learning algorithms.
Artificial Intelligence (AI) has become an integral part of our daily lives and recently gained widespread public interest with the impressive introduction of ChatGPT in November 2022. The popular chatbot utilizes the technology of Deep Neural Networks (DNNs), which has become popular among researchers since the network “AlexNet” scored first place in the ImageNet competition in 2012. Despite their outstanding performance, DNNs poses significant technical and environmental challenges. The computational and memory requirements during training and inference are immense, resulting in substantial energy consumption and carbon dioxide emissions. In addition, high-performance computing hardware, such as a GPU, is necessary to achieve a reasonable runtime, preventing DNNs from running directly on mobile or embedded devices.
Network quantization is a promising way to solve these challenges. Instead of using 32-bit floating-point values, lower-bit values are used to represent features and weights in the network. The most extreme case, a Binary Neural Network (BNN), uses 1-bit values, which allows for model compression and speedup during inference, albeit at the cost of slightly decreased accuracy. The efficient implementation allows these binary models to run on less-powerful CPUs and preserve energy simultaneously, e.g.., for running on battery-powered devices.
This thesis explores various model design principles that enhance the accuracy and efficiency of BNNs and demonstrates their effectiveness on standard benchmark datasets. Before proposing our first model architecture and approach, we examine the importance of shortcut connections in previous BNN architectures and formulate golden rules for designing accurate and compact BNNs. These insights are then used to remove network bottlenecks and construct BinaryDenseNet based on dense shortcut connections.
In the second approach, we concentrate on solving both core problems of binary networks: reduced feature capacity and reduced feature quality at the same time. The proposed solution, combining dense blocks with a novel improvement block design, is the basis for building the network MeliusNet. The evaluation of MeliusNet shows it can challenge the accuracy and efficiency of the popular compact neural network MobileNet.
Finally, we concentrate on maximizing the energy efficiency of BNNs by reducing the remaining 32-bit values and propose BoolNet in our third approach. The energy consumption is measured through hardware simulations and shows a more efficient result than other works.
To promote reproducibility and support future research, the code for all proposed methods in this thesis is published individually or as part of open-source frameworks developed during this work. The technical design of these frameworks is presented briefly in the later parts of this work, together with demo applications. Afterward, we present the related work to the different approaches in this work and conclude with an outlook to the future of BNNs.
Human nutrition relies on crop yield, which is affected by environmental conditions, such as the availability of nutrients and water, light intensity as well as temperature. Climate change is projected to severely impact crop yield due to more fluctuating and increasing temperatures. Metabolism, the life-sustaining network of biochemical reactions in a cell that transforms nutrients into the building blocks of biomass, is the most immediate level of response to changes in extrinsic factors, like temperature. Constraint-based modeling of metabolic networks offers the means to predict environment-specific metabolic phenotypes including: growth rate, individual reaction fluxes, and protein abundances, driven by the recent developments of enzyme-constraint metabolic models (ecGEMs). This modeling framework has been instrumental in identifying factors that limit growth under (sub)optimal environmental conditions and guiding metabolic engineering across different species.
The extent to which metabolism can adjust in response to environmental changes is determined by the ability of flux rerouting through the metabolic network, termed metabolic plasticity. This thesis comprises three studies that investigate the metabolic plasticity of the model plant Arabidopsis thaliana to changes in environmental conditions as well as an approach for the improvement of ecGEM parameterization that is yet to be applied in plants. In the first study, the re-routing of reaction fluxes in photorespiratory mutants of A. thaliana was investigated under constant and fluctuating light conditions using thermodynamics-based metabolic flux analysis. This was the first time the method was applied in plants. Differences in growth of the mutants between constant and fluctuating light conditions could be explained by reduced photosynthesis rates and flux re-routing. The second study presents a constraint-based approach that improves the estimates of enzyme turnover numbers (kcat) in ecGEMs by combining proteomic and physiological data from multiple growth conditions. In the third study, an ecGEM of A. thaliana metabolism was created and coupled with temperature-dependent constraints on enzyme kinetics, total protein content, and photosynthesis. The resulting model was in turn used to predict relative growth rates and net CO2 assimilation rates at different temperatures, in agreement with experimental data. The model allowed us to show that metabolic flexibility decreases with increasing temperature and to identify metabolic pathways with differential responses to temperature change. Finally, predictions of growth reductions in knock-out lines at 17 °C were tested and validated by dry weight measurements of T-DNA lines in a temperature shift experiment. Therefore, this thesis paves the way for identifying complex management and engineering strategies using mechanistic modeling approaches that consider the influence of abiotic factors on plant growth and metabolism.
The study initiates with the properties of triple cation pin solar cells as a function of their changing bandgap. The standard device layout comprises PTAA as hole-transport layer (HTL) and C60/BCP as electron-transport layer (ETL). The bandgap of the composition Cs0.05(FAxMA1-x)0.95Pb(IxBr1-x)3 can be linearly changed from 1.52 eV (x=1) to 1.88 eV (x=0.5). However, the VOC of the devices plateaus around 1.17 V instead of increasing as theoretically expected. Although these VOC losses were traditionally ascribed to halide segregation, we quantified the radiative efficiency losses of the devices by measuring the electroluminescence quantum efficiency Qlume as a function of the bromine content. The VOC calculated from Qlume matches the VOC of the J-V measurements in all the cases, concluding that non-radiative recombination losses are directly proportional to the bromine content, hence the VOC pinning. Non-radiative losses can be partially mitigated by the addition of LiF as a passivation layer before the ETL until the bandgap reaches 1.74 eV. Above this value, both the n and p interfaces require further optimisation to increase the VOC. Changing the HTL to 2PACz, adding oleylamine to the precursor solution and LiF as a passivation agent before the C60, accomplish altogether a stepwise reduction in the non-radiative losses attaining finally 83% of the radiative limit for the open-circuit voltage (VOC,rad) in cells with bandgaps of 1.80, 1.85 and 1.88 eV. Subsequently, we fabricated monolithic 2-terminal tandem solar cells with a Pb-Sn perovskite as the low-bandgap subcell and a recombination layer of AZO/SnOx/InOx. We attempted the three already-optimised aforementioned bandgaps (1.80, 1.85 and 1.88 eV) for the top high-bandgap subcell. The best combination was found to be 1.85/1.27 eV reaching 23.7% stabilised power output. Further analysis included subcell characterisation by (injection dependent) electro- and (intensity dependent) photoluminescence to assess both the implied efficiency potential and the limitations. Transport losses seen in the FF reduce the performance of the tandem from its efficiency potential of 25.2%. The low external quantum efficiency seen in the low-bandgap standard subcell is explained but its non-optimal absorption, an issue that is solved by increasing the thickness of the active layer, boosting in that way the efficiency up to 25.9% in a proof-of-concept device with a 30% of implied efficiency. In order to gain insights in the device physics, we investigated several perovskite compositions to quantify the electronic doping and its impact on device performance. Using AC Hall effect, this doping density is of the order of 1e11-1e13 cm-3 in Pb-based perovskites. By observing how the recombination regimes change upon the light intensity in photoluminescence quantum yield and transient photoluminescence measurements, it was found that Sn-based perovskites exhibit a higher doping concentration (1e14 cm-3) in comparison to Pb-based ones (1e12 cm-3). By using charge-extraction techniques, the electrode charge per cell volume under short-circuit conditions (1e16 cm-3), was found to be much higher than the presumable doping density, concluding (supported by simulations) that the latter does not affect the device performance. However, at longer time scales (ms-s) much larger charge concentrations were detected. We concluded that the changes in the internal electrostatic field do not come from doping but from these concentration of mobile ions that amount to roughly 1e17 cm-3 in fresh devices. This ionic concentration can be assessed and accurately quantified by bias-assisted charge extraction (BACE) and by measuring the J-V scans of the cells at different scan rates. At high scan speeds, the ions do not react to the changing field and therefore, neither field screening nor hysteresis are observed. At low scan speeds, hysteresis is not present either, yet the ions have enough time to go back to interface where they screen the internal field the most. In the points in between, both hysteresis and field screening manifest depending on the scan rate. The accuracy of charge extraction under linearly increasing voltage (CELIV) in comparison to BACE when it comes to the determination of the ionic density is lower owing to the initial ionic distribution at short-circuit condition. Mott-Schottky analysis at low frequency was performed in order to complement BACE in the estimation of the ionic density, which result in a value of 2.4e17 cm-3, which rises to 7.5e17 cm-3 after 5 hours under 1 sun equivalent illumination and open-circuit conditions.
Networks are widely used across many fields to represent interactions, such as the web, social interactions, infrastructure, or biological connections. Algorithms and processes on these networks have many uses, for example to find the shortest path in a road network, or to understand the spreading of information through a social network. In order to better answer such questions, we should understand how algorithms and processes are affected by features of the networks. One useful tool for this can be found in random network models, serving as proxies for real-world networks. Their features can be controlled via model parameters. While these models are usually easier to understand and analyze than real-world networks, their usefulness depends on how realistic they are.
Thus, we look at well-known network models and evaluate whether they are useful for predicting algorithm behavior on real-world networks - the external validity. We focus on many widely-used algorithms and processes, and establish that the models are good proxies for real-world networks. We observe that heterogeneity and locality are important features for explaining and predicting algorithm behavior.
In order to use random network models as proxies for real-world networks or to investigate model fit, one should utilize the model configuration capabilities optimally. As the models can be configured via parameters that affect the network features, this entails choosing the best parameters for some target network features. However, the exact relation between the configuration and the resulting network features is not clear. We present a method for estimating the best-fitting parameter configuration for commonly used network models, given some target features of the resulting networks. Our iterative method based on stochastic approximation needs only few samples and works reliably across a range of parameter configurations.
Furthermore, we take a closer look at the features and substructures of networks. In particular, based on methods from clustering and outlier detection, we consider the idea that the edge set actually comes from two edge sets with differing features. With this idea, we investigate the well-known bootstrap percolation process on this combination of networks, which had previously only been considered on single network models. We theoretically and empirically consider the process on two combined network models, with differing percolation thresholds for the two edge sets, and see an interesting emerging behavior of a slow and a rapid percolation phase. In addition, we consider ways to separate edge sets based on locality features. Assuming the edges are a combination of two network models, we show how the edges can be separated by a simple metric, even with imbalanced set sizes.
Overall, we evaluate common random network models, improve methods to utilize them, and contribute towards the design and understanding of new, combined network models.
Visualising Southern African Late Iron Age Settlements in the Digital Age studies the visualisation of Southern African Late Iron Age Settlements (LIAS) (c. 900–1800) across the late nineteenth, twentieth, and early twenty-first centuries (1871–2020), as found in a survey of the cultural production, circulation, reproduction, and theorisation of illustrations accompanying archaeological, anthropological, and historical Southern African LIAS research. A valuable contribution of LIAS research is its continuous demonstration of a pre-colonial hub of cosmopolitanisms on a scale never imagined in colonial histories of 'indigenous' communities – thought of as the ultimate 'other' of global modernity.
This study focuses on the visualisation of four settlements, namely: Mapungubwe, Khami, Great Zimbabwe, and Bokoni. It is proposed that as with the authority of Eurocentric 'formative interpretations' of LIAS research currently under review, visualisations accompanying LIAS also need to be critically relooked at within appropriate visual cultural methodologies informed by postcolonial, decolonial and critical race theory. The study follows a two-fold methodological framework involving a textual analysis and an image-making process. On both accounts, the study focuses on the cultural politics of representation, asking: who and what is being made visible in the visualisation of settlements accompanying LIAS research; what forms of materiality and spatiality are pictured and performed; what is the affect such visualisations have on the people that experience them; and finally, what do they mean in the context in which they are made?
The Arctic climate is currently experiencing a notable rise in air temperatures and an altered precipitation dynamic. These changes are not just regional phenomena; they have far-reaching implications for the global climate. Temperature records from the past few decades unequivocally show that the Arctic is warming at a rate of more than four times the global average, a phenomenon known as Arctic amplification. This warming trend has led to a significant reduction in sea-ice extent and thickness, fundamentally altering the region’s albedo effect, its ability to reflect solar radiation back into space. As sea-ice diminishes, dark ocean waters absorb more sunlight, further warming the region and accelerating ice melt. This dissertation delves into these transformations, focusing on the characteristics of stable water isotopes in the Arctic water cycle and their utility in understanding the multifaceted connections between atmosphere, ocean, and sea-ice.
Central to this dissertation is the utilization of stable water isotopes as a diagnostic proxy to unravel the dynamics of the Arctic hydrosphere and cryosphere. Through the comprehensive analysis of over 2200 samples of seawater, snow, and sea-ice collected during the Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) expedition, this work provides an unprecedented view of the isotopic signatures of the components shaping the central Arctic water cycle. The study highlights how the isotopic composition of sea-ice varies, revealing insights into its formation processes. First-year ice (FYI) is generally more enriched in δ18O than second-year ice (SYI), due to its initial freezing from the seawater with enriched δ18O, whereas SYI is influenced by contributions of snow melt. A distinct isotopic threshold between FYI and SYI, helped characterizing an insulated FYI (iFYI) layer underneath the SYI. Hence, different sources and formation mechanisms of sea ice could be differentiated. Furthermore, the isotopic signatures of surface seawater offer insights into spatio-temporal patterns of relative contributions of different freshwater sources from Siberian rivers to Greenland Ice Sheet melt as well as summer sea-ice and snow melt to the Arctic Ocean’s water budget.
A ground-breaking discovery in this dissertation using the MOSAiC expedition data is that a significant portion, approximately 20 %, of the Arctic’s winter snow cover originates not from traditional meteoric sources, but directly from vapour diffusion from the sea-ice through sublimation. This process occurs under the intense temperature gradients characteristic of Arctic winters, contributing to the formation of a unique snow-like structure atop the sea-ice. This “oceanic” source of snow introduces new perspectives on understanding vapour fluxes across the snowpack, the biogeochemistry of gas exchanges, and the implications for sea salt aerosol formation. The implications extend further, affecting mass-balance calculations and physical properties of snow on Arctic sea-ice.
Over a year-long period, the investigation of Arctic snow highlights significant seasonal changes closely tied to local or regional atmospheric conditions, based on 911 snow isotope measurements. Snow, with its highly variable and depleted δ18O values, plays a crucial role in the Arctic water cycle. During autumn, an inherited signal in snow isotopes from earlier precipitation events led to distinct differences between δ18O values of snow and water vapour. During the winter months, when the RV Polarstern transitioned from the Siberian to the Atlantic sector of the Arctic Ocean, significant differences in δ18O and d-excess values in snow and vapour were noted. These differences suggest kinetic fractionation, likely primarily driven by sublimation, during the severe cold and dry conditions. This observation is associated with a generally low statistical correlation between the δ18O in snow and air temperature. This highlights the greater influence of post-depositional processes on the snow isotopes, compared to the those during deposition. Wind-driven snow re-distribution, occurring consistently throughout the winter, led to a mixed but depleted δ18O signal in surface snow across the sea-ice by spreading meteoric snow of lower δ18O. This effect is more evident in ridge situations, contrasting sharply with flat ice samples in both, snow profile heights and isotopic compositions. Summer months show isotopic alignment between surface snow and vapour under warmer conditions, suggesting equilibrium fractionation between them.
In general, this dissertation contributes to a better understanding of the Arctic water cycle and their stable isotopic signatures, emphasizing their importance in deciphering the feedback mechanisms driving current changes in the Arctic climate. By offering detailed insights into the interactions between snow, sea-ice, ocean and atmosphere, this work advances our knowledge of the Arctic system.