This thesis is dedicated to the study of the turbulent fluid flow in rough-walled rock fractures. Fracture models were generated from 3D scans of fractured rock samples, while fluid flow was simulated numerically by means of FVM-based open-source CFD toolbox OpenFOAM, employing the high-performance computing cluster for the more demanding 3D models. First part of the thesis addresses the issue of fracture geometry. Realistic 2D and 3D fracture models were constructed from 3D scans of upper and lower halves of a fractured rock sample, taking both shear displacement and contact spots into account. Furthermore, we discuss the shortcomings of the available fracture aperture metrics and propose a new aperture metric based on the Hausdorff distance; imaging performance of the new metric is shown to be superior to the conventional vertical aperture, especially for rough fracture surfaces with abundant ridges and troughs. In the second part of the thesis we focus on the fluid flow through the rock fracture for both 2D and 3D cases. While previous studies were largely limited to the fully viscous Darcy or inertial Forchheimer laminar flow regimes, we chose to investigate across the widest possible range of Reynolds numbers from 0.1 to 10^6, covering both laminar and turbulent regimes, which called for a thorough investigation of suitable turbulence modeling techniques. Due to narrow mean aperture and high aspect ratio of the typical fracture geometry, meshing posed a particularly challenging problem. Taking into account limited computational resources and a sheer number of model geometries, we developed a highly-optimised workflow, employing the steady-state RANS simulation approach to obtain time-averaged flow fields. Our findings show that while flow fields remain mostly stationary and undisturbed for simpler contactless geometries, emergence of contact spots immediately triggers a transition to non-stationary flow starting from Re ∼ 10^2, which is reflected by the streamline tortuosity data. This transition disrupts the flow pattern across the fracture plane, causing strong channeling and large separation bubbles, with area of the latter being much larger than the generating contact spots. Adverse influence of the contact spots on the overall permeability is strong enough to override any benefits of aperture increase during shear and dilation. Contactless 3D models can to a certain degree be approximated by their 2D counterparts. Lastly, we investigate the influence of both shearing and contact spots on the overall permeability and friction factor of the fracture, drawing a parallel to the well-studied area of turbulent flow in rough-walled pipes and ducts. Unlike the latter, 3D curvilinear fracture geometries exhibit a gapless laminar–turbulent transition, behaving as a hydraulically rough channel in the turbulent range as the shear displacement increases.

Water resources management has to deal with complex real life problems under uncertain framework conditions. One possibility for encountering such challenges is integrated water resources management (IWRM). However, IWRM is often understood as prescriptive manual, not acknowledging the need for adaptive solutions and capacity development (CD). These challenges demonstrate that sustainable water resources management requires transdisciplinarity, i.e. the integration of several scientific disciplines, as well as the collaboration between science and local actors. Transdisciplinarity is inherently related to CD since it facilitates collaboration and provides mutual learning and knowledge on complex interrelationships. This correlates with the evidence that CD can be seen as a key factor for water resources management (Alaerts et al. 1991, Alaerts 2009). Consequently, the objective of this thesis is to strengthen water resources management by connecting processes of IWRM and CD in a transdisciplinary sense, i.e. (i) interrelating disciplinary research within an interdisciplinary research team that collaborates with local actors, and (ii) conducting a political process for knowledge and capacity development. Based on general insights, an embedded case study in the Western Bug River Basin, Ukraine, was conducted to evaluate the concept. It is shown that CD is essential for shifting from IWRM theories towards implementation and accordingly advantages of harmonizing CD into the IWRM process are presented (Leidel et al. 2012). Next to capacity issues, also other coordination gaps were assessed. River Basin Organisations are frequently proposed as a response to the administrative gap; however, coordination efforts cannot be simply reduced by transferring tasks from jurisdictional institutions to a river basin authority, because they will always need to coordinate with organizations from within or outside the water sector (von Keitz and Kessler 2008). Thus, coordination mechanisms across the boundaries of relevant policy fields are essential. Therefore, a management framework is established linking technical development and capacity development that describes interrelations between environmental pressures and capacity and information gaps for different levels of water management (Leidel et al. 2014). The developed model-based and capacity-based IWRM framework combines model-based systems analysis and capacity analysis for developing management options that support water management actors. This is aligned with a political process for capacity development. It constitutes a boundary object for approaching cross-scale challenges that converges analyses, assessments and participation into one strategy. As concluded by Mollinga (2008), this can improve the performance of sustainable resources management by approaching transdisciplinarity. Within the model and capacity-based IWRM framework, the results of the integrated analysis are made explicit and transparent by introducing a matrix approach. Technical issues, institutional challenges, organizational and human resources development, and information needs are jointly assessed and interrelated by confronting pressures and coordination gaps on a subsystem basis. Accordingly, the concept supports a transparent decision making process by identifying knowledge and capacities required for the implementation of technical intervention options and vice versa. The method is applied in the International Water Research Alliance Saxony (IWAS) model region ‘Ukraine’. It could be shown that the approach delivers management options that are scientifically credible and also accepted by and relevant for the actors. The case study revealed that technical intervention measures for the urban and rural water management have to be jointly implemented with appropriate CD measures and an accompanying political process on (i) strengthening the institutional framework and interministerial collaboration, (ii) fitting RBM into the existing institutional framework, (iii) setting up prerequisites for realistic RBM (Monitoring, information management, legal enforcement), (iv) a revision of effluent standards and a differentiated levy system, (v) cost covering tariffs, (vi) association work. For the Western Bug River Basin (WBRB), the strengthening of the collaboration between actors on all levels has to be continued. For increasing the usability, the approach needs to be institutionalized and become more practice relevant, e.g. by extending it to a water knowledge management system. Developing a roadmap for establishing transboundary water management is a subsequent step. For strengthening future water management actors, IWRM curricula development at uni-versities in Ukraine was supported. And we developed the e-learning module IWRM-education that links interactively different aspects of water management to comprehend the complexity of IWRM (Leidel et al. 2013). The evaluation showed that participants under-stand the content, appreciate this way of learning, and will use this module for further activities. The case study showed that technical cooperation can be a facilitator for political processes and that it can support decision making in a transparent way. Yet, it also showed that IWRM is highly political process and that the developed approach cannot cover all obstacles. In summary, exploring and reducing simultaneously environmental pressures and capacity and information gaps is essential for water sector evolution worldwide. Accordingly, transdisciplinarity as a means for capacity development can support the implementation of real integrated water resources management.

Zum Wintersemester 2019/2020 wird an der TU Bergakademie Freiberg ein neuer MSc- Studiengang „Geomatics for Mineral Resource Management“ angeboten. Dieser ist Teil eines internationalen Programmes unter Beteiligung der Universitäten Técnico Lisboa in Portugal, Delft University of Technology in den Niederlanden, TU Bergakademie Freiberg in Deutschland, Montanuniversität Leoben in Österreich sowie Wroclaw University of Science and Technology in Polen. Es ist vorgesehen, ein europaweit sichtbares Programm anzubieten, dass den Studierenden ein flexibles internationales Studium an jeweils zwei der Hochschulen ermöglicht und sie auf Führungsaufgaben im Bereich der Geomatik in der Rohstoffwirtschaft vorbereitet. Der Beitrag fasst das Konzept des internationalen Programmes zusammen und stellt die Möglichkeiten an der TU Freiberg konkret dar. / For the winter term 2019/2020, a new MSc program 'Geomatics for Mineral Resource Management' will be offered at the TU Bergakademie Freiberg. This is part of an international program involving the Universities of Técnico Lisboa in Portugal, Delft University of Technology in the Netherlands, TU Bergakademie Freiberg in Germany, Montanuniversität Leoben in Austria and Wroclaw University of Science and Technology in Poland. The ambition is to offer a Europe-wide visible program that enables students to study internationally flexibly at two of the universities being prepared for a leader role in geomatics for Mineral Resource Management. The article summarizes the concept of the international program and presents the possibilities at the TU Freiberg.

When specimens of different sands are produced using the same preparation method and sheared under the same conditions (consolidation stress, loading, etc.), while simultaneously keeping the drainage closed, the resulting tendencies of these sands regarding the PWP build-up will be different. This research paper presents a simplified cyclic shear test, which is used to evaluate the accumulation of PWP in sands under defined specimen preparation procedure and testing conditions. In the proposed experiment, a comparison of different sands with this respect is easily achieved. The principle of this experimental method is based on the evolution of the PWP during cyclic shearing of a water-saturated sand sample. Undrained conditions during the experiment allow for the evolution of the PWP, which is quantified by the rate of the PWP build-up. The duration of a single cyclic shear test, including specimen preparation, is approximately 30 min. The evaluation of the rate of the PWP build-up for different densities resulted in an exponential dependence of the PWP build-up on the variation of the relative density. The results confirmed a higher generation of PWP in a fine sand compared to a coarse sand. A comparison with the results of undrained cyclic triaxial tests in the case of eight different sands demonstrated a good agreement between both experimental methods. The basis for the comparison was the density-dependent evolution of PWP in these methods. The presented method delivers a value (index) that quantifies the PWP build-up in sands under the defined testing conditions.

Computed Tomography (CT) imaging provides invaluable insight into internal structures of objects and organisms, which is critical for applications ranging from materials science to medical diagnostics. In CT data, an object is represented by a 3D reconstruction that is generated by combining multiple 2D X-ray images taken from various angles around the object. Each voxel, a volumetric pixel, within the reconstructed volume represents a small cubic element, allowing for detailed spatial representation. To extract meaningful information from CT imaging data and facilitate analysis and interpretation, accurate segmentation of internal structures is essential. However, this can be challenging due to various artifacts introduced by the physics of a CT scan and the properties of the object being imaged. This dissertation directly addresses this challenge by using deep learning techniques. Specifically, Convolutional Neural Networks (CNNs) are used for segmentation. However, they face the problem of limited training data. Data scarcity is addressed by data augmentation through the unsupervised generation of synthetic training data and the use of 2D and 3D data augmentation methods. A combination of these augmentation strategies allows for streamlining segmentation in voxel data and effectively addresses data scarcity. Essentially, the work aims to simplify training of CNNs, using minimal or no labeled data. To enhance accessibility to the results of this thesis, two user-friendly software solutions, unpAIred and AiSeg, have been developed. These platforms enable the generation of training data, data augmentation, as well as training, analysis, and application of CNNs. This cumulative work first examines simpler but efficient conventional data augmentation methods, such as radiometric and geometric image manipulations, which are already widely used in literature. However, these methods are usually randomly applied and do not follow a specific order. The primary focus of the first paper is to investigate this approach and to develop both online and offline data augmentation pipelines that allow for systematic sequencing of these operations. Offline augmentation involves augmenting training data stored on a drive, while online augmentation is performed dynamically at runtime, just before images are fed to the CNN. It is successfully shown that random data augmentation methods are inferior to the new pipelines. A careful comparison of 3D CNNs is then performed to identify optimal models for specific segmentation tasks, such as carbon and pore segmentation in CT scans of Carbon Reinforced Concrete (CRC). Through an evaluation of eight 3D CNN models on six datasets, tailored recommendations are provided for selecting the most effective model based on dataset characteristics. The analysis highlights the consistent performance of the 3D U-Net, one of the CNNs, and its residual variant, which excel at roving (a bundle of carbon fibers) and pore segmentation tasks. Based on the augmentation pipelines and the results of the 3D CNN comparison, the pipelines are extended to 3D, specifically targeting the segmentation of carbon in CT scans of CRC. A comparative analysis of different 3D augmentation strategies, including both offline and online augmentation variants, provides insight into their effectiveness. While offline augmentation results in fewer artifacts, it can only segment rovings already present in the training data, while online augmentation is essential for effectively segmenting different types of rovings contained in CT scans. However, constraints such as limited diversity of the dataset and overly aggressive augmentation that resulted in segmentation artifacts require further investigation to address data scarcity. Recognizing the need for a larger and more diverse dataset, this thesis extends the results of the three former papers by introducing a deep learning-based augmentation using a Generative Adversarial Network (GAN), called Contrastive Unpaired Translation (CUT), for synthetic training data generation. By combining the GAN with augmentation pipelines, semi-supervised and unsupervised end-to-end training methods are introduced and the successful generation of training data for 2D pore segmentation is demonstrated. However, challenges remain in achieving a stable 3D CUT implementation, which warrants further research and development efforts. In summary, the results of this dissertation address the challenges of accurate CT data segmentation in materials science through deep learning techniques and novel 2D and 3D online and offline augmentation pipelines. By evaluating different 3D CNN models, tailored recommendations for specific segmentation tasks are provided. Furthermore, the exploration of deep learning-based augmentation using CUT shows promising results in the generating synthetic training data. Future work will include the development of a stable implementation of a 3D CUT version, the exploration of new model architectures, and the development of sub-voxel accurate segmentation techniques. These have the potential for significant advances in segmentation in tomography data.:Abstract IV Zusammenfassung VI 1 Introduction 1 1.1 Thesis Structure 2 1.2 Scientific Context 3 1.2.1 Developments in the Segmentation in Tomography Data 3 1.2.2 3D Semantic Segmentation using Machine Learning 5 1.2.3 Data Augmentation 6 2 Developed Software Solutions: AiSeg and unpAIred 9 2.1 Software Design 10 2.2 Installation 11 2.3 AiSeg 11 2.4 unpAIred 12 2.5 Limitations 12 3 Factors Affecting Image Quality in Computed Tomography 13 3.1 From CT Scan to Reconstruction 13 3.2 X-ray Tube and Focal Spot 14 3.3 Beam Hardening 14 3.4 Absorption, Scattering and Pairing 15 3.5 X-ray Detector 16 3.6 Geometric Calibration 17 3.7 Reconstruction Algorithm 17 3.8 Artifact corrections 18 4 On the Development of Augmentation Pipelines for Image Segmentation 19 4.0 Abstract 20 4.1 Introduction 20 4.2 Methods 21 4.2.1 Data Preparation 21 4.2.2 Augmentation 21 4.2.3 Networks 24 4.2.4 Training and Metrics 25 4.3 Experimental Design 26 4.3.1 Hardware 26 4.3.2 Workflow 26 4.3.3 Test on Cityscapes 26 4.4 Results and Discussion 26 4.4.1 Stage 1: Crating a Baseline 27 4.4.2 Stage 2: Using Offline Augmentation 27 4.4.3 Stage 3: Using Online Augmentation 27 4.4.4 Test on Cityscapes 29 4.4.5 Future Work – A New Online Augmentation 30 4.5 Conclusion 31 4.6 Appendix 31 4.6.1 Appendix A. List of All Networks 31 4.6.2 Appendix B. Augmentation Methods 32 4.6.3 Appendix C. Used RIWA Online Augmentation Parameters 36 4.6.4 Appendix D. Used Cityscapes Online Augmentation Parameters 36 4.6.5 Appendix E. Comparison of CNNs with best Backbones on RIWA 37 4.6.6 Appendix F. Segmentation Results 38 4.7 References 39 5 Comparison of 3D CNNs for Volume Segmentation 43 5.0 Abstract 44 5.1 Introduction 44 5.2 Datasets 44 5.2.1 Carbon Rovings 45 5.2.2 Concrete Pores 45 5.2.3 Polyethylene Fibers 45 5.2.4 Brain Mitochondria 45 5.2.5 Brain Tumor Segmentation Challenge (BraTS) 46 5.2.6 Head and Neck Cancer 46 5.3 Methods 46 5.3.1 Data Preprocessing 46 5.3.2 Hyperparameters 46 5.3.3 Metrics 47 5.3.4 Experimental Design 48 5.4 Results and Discussion 48 5.4.1 Impact of Initial Random States (Head and Neck Cancer Dataset) 48 5.4.2 Carbon Rovings 48 5.4.3 Concrete Pores 49 5.4.4 Polyethylene Fibers 49 5.4.5 Brain Mitochondria 50 5.4.6 BraTS 51 5.5 Conclusion 51 5.6 References 52 6 Segmentation of Carbon in CRC Using 3D Augmentation 55 6.0 Abstract 56 6.1 Introduction 56 6.2 Materials and Methods 58 6.2.1 Specimens 58 6.2.2 Microtomography 59 6.2.3 AI-Based Segmentation 60 6.2.4 Roving Extraction 64 6.2.5 Multiscale Modeling 65 6.2.6 Scaled Boundary Isogeometric Analysis 66 6.2.7 Parameterized RVE and Definition of Characteristic Geometric Properties 67 6.3 Results and Discussion 70 6.3.1 Microtomography 70 6.3.2 Deep Learning 71 6.3.3 Roving Extraction 74 6.3.4 Parameterized RVE and Definition of Characteristic Geometric Properties 75 6.4 Conclusion 79 6.5 References 80 7 Image-to-Image Translation for Semi-Supervised Semantic Segmentation 85 7.1 Introduction 85 7.2 Methods 86 7.2.1 Generative Adversarial Networks 87 7.2.2 Contrastive Unpaired Translation 87 7.2.3 Fréchet Inception Distance 89 7.2.4 Datasets 89 7.3 Experimental Design 92 7.4 Results and Discussion 94 7.4.1 Training and Inference of CUT 94 7.4.2 End-to-End Training for Semantic Segmentation 99 7.5 Conclusion 104 7.5.1 Future Work 104 8 Synthesis 107 8.1 Research Summary 107 8.1.1 Augmentation Pipelines 107 8.1.2 3D CNN Comparison 108 8.1.3 3D Data Augmentation for the Segmentation of Carbon Rovings 108 8.1.4 Synthetic Training Data Generation 109 8.2 Future Developments 109 8.2.1 Augmentation 109 8.2.2 Pre-trained 3D Encoder 111 8.2.3 On the Quality Control of Carbon Reinforced Concrete 111 8.2.4 Subvoxel Accurate Segmentation 113 8.2.5 Towards Volume-to-Volume Translation 114 8.3 Conclusion 114 References 117 List of Tables 125 List of Figures 127 List of Abbreviations 131 / Computertomographie (CT) bietet wertvolle Einblicke in die inneren Strukturen von Objekten und Organismen, was für Anwendungen von der Materialwissenschaft bis zur medizinischen Diagnostik von entscheidender Bedeutung ist. In CT-Daten ist ein Objekt durch eine 3D-Rekonstruktion dargestellt, die durch die Kombination mehrerer 2D-Röntgenbilder aus verschiedenen Winkeln um das Objekt herum erstellt wird. Jedes Voxel, ein Volumen Pixel, innerhalb des rekonstruierten Volumens stellt ein kleines kubisches Element dar und ermöglicht eine detaillierte räumliche Darstellung. Um aussagekräftige Informationen aus CT-Bilddaten zu extrahieren und eine Analyse und Interpretation zu ermöglichen, ist eine genaue Segmentierung der inneren Strukturen unerlässlich. Dies kann jedoch aufgrund verschiedener Artefakte, die durch die Physik eines CT-Scans und Eigenschaften des abgebildeten Objekts verursacht werden, eine Herausforderung darstellen. Diese Dissertation befasst sich direkt mit dieser Herausforderung, indem sie Techniken des Deep Learnings einsetzt. Konkret werden für die Segmentierung Convolutional Neural Networks (CNNs) verwendet, welche jedoch mit dem Problem begrenzter Trainingsdaten konfrontiert sind. Der Datenknappheit wird dabei durch Datenerweiterung begegnet, indem unbeaufsichtigt synthetische Trainingsdaten erzeugt und 2D- und 3D-Augmentierungssmethoden eingesetzt werden. Eine Kombination dieser Vervielfältigungsstrategien erlaubt eine Vereinfachung der Segmentierung in Voxeldaten und behebt effektiv die Datenknappheit. Im Wesentlichen zielt diese Arbeit darauf ab, das Training von CNNs zu vereinfachen, wobei wenige oder gar keine gelabelten Daten benötigt werden. Um die Ergebnisse dieser Arbeit Forschenden zugänglicher zu machen, wurden zwei benutzerfreundliche Softwarelösungen, unpAIred und AiSeg, entwickelt. Diese ermöglichen die Generierung von Trainingsdaten, die Augmentierung sowie das Training, die Analyse und die Anwendung von CNNs. In dieser kumulativen Arbeit werden zunächst einfachere, aber effiziente konventionelle Methoden zur Datenvervielfältigung untersucht, wie z. B. radiometrische und geometrische Bildmanipulationen, die bereits häufig in der Literatur verwendet werden. Diese Methoden werden jedoch in der Regel zufällig nacheinander angewandt und folgen keiner bestimmten Reihenfolge. Der Schwerpunkt des ersten Forschungsartikels liegt darin, diesen Ansatz zu untersuchen und sowohl Online- als auch Offline-Datenerweiterungspipelines zu entwickeln, die eine systematische Sequenzierung dieser Operationen ermöglichen. Bei der Offline Variante werden die auf der Festplatte gespeicherten Trainingsdaten vervielfältigt, während die Online-Erweiterung dynamisch zur Laufzeit erfolgt, kurz bevor die Bilder dem CNN gezeigt werden. Es wird erfolgreich gezeigt, dass eine zufällige Verkettung von geometrischen und radiometrischen Methoden den neuen Pipelines unterlegen ist. Anschließend wird ein Vergleich von 3D-CNNs durchgeführt, um die optimalen Modelle für Segmentierungsaufgaben zu identifizieren, wie z.B. die Segmentierung von Carbonbewehrung und Luftporen in CT-Scans von carbonverstärktem Beton (CRC). Durch die Bewertung von acht 3D-CNN-Modellen auf sechs Datensätzen werden Empfehlungen für die Auswahl des genauesten Modells auf der Grundlage der Datensatzeigenschaften gegeben. Die Analyse unterstreicht die konstante Überlegenheit des 3D UNets, eines der CNNs, und seiner Residualversion bei Segmentierung von Rovings (Carbonfaserbündel) und Poren. Aufbauend auf den 2D Augmentierungspipelines und den Ergebnissen des 3D-CNN-Vergleichs werden die Pipelines auf die dritte Dimension erweitert, um insbesondere die Segmentierung der Carbonbewehrung in CT-Scans von CRC zu ermöglichen. Eine vergleichende Analyse verschiedener 3D Augmentierungsstrategien, die sowohl Offline- als auch Online-Erweiterungsvarianten umfassen, gibt Aufschluss über deren Effektivität. Die Offline-Augmentierung führt zwar zu weniger Artefakten, kann aber nur Rovings segmentieren, die bereits in den Trainingsdaten vorhanden sind. Die Online-Augmentierung erweist sich hingegen als unerlässlich für die effektive Segmentierung von Carbon-Roving-Typen, die nicht im Datensatz enthalten sind. Einschränkungen wie die geringe Vielfalt des Datensatzes und eine zu aggressive Online-Datenerweiterung, die zu Segmentierungsartefakten führt, erfordern jedoch weitere Methoden, um die Datenknappheit zu beheben. In Anbetracht der Notwendigkeit eines größeren und vielfältigeren Datensatzes erweitert diese Arbeit die Ergebnisse der drei Forschungsartikel durch die Einführung einer auf Deep Learning basierenden Augmentierung, die ein Generative Adversarial Network (GAN), genannt Contrastive Unpaired Translation (CUT), zur Erzeugung synthetischer Trainingsdaten verwendet. Durch die Kombination des GANs mit den Augmentierungspipelines wird eine halbüberwachte Ende-zu-Ende-Trainingsmethode vorgestellt und die erfolgreiche Erzeugung von Trainingsdaten für die 2D-Porensegmentierung demonstriert. Es bestehen jedoch noch Herausforderungen bei der Implementierung einer stabilen 3D-CUT-Version, was weitere Forschungs- und Entwicklungsanstrengungen erfordert. Zusammenfassend adressieren die Ergebnisse dieser Dissertation Herausforderungen der CT-Datensegmentierung in der Materialwissenschaft, die durch Deep-Learning-Techniken und neuartige 2D- und 3D-Online- und Offline-Augmentierungspipelines gelöst werden. Durch die Evaluierung verschiedener 3D-CNN-Modelle werden maßgeschneiderte Empfehlungen für spezifische Segmentierungsaufgaben gegeben. Darüber hinaus zeigen Untersuchungen zur Deep Learning basierten Augmentierung mit CUT vielversprechende Ergebnisse bei der Generierung synthetischer Trainingsdaten. Zukünftige Arbeiten umfassen die Entwicklung einer stabilen Implementierung einer 3D-CUT-Version, die Erforschung neuer Modellarchitekturen und die Entwicklung von subvoxelgenauen Segmentierungstechniken. Diese haben das Potenzial für bedeutende Fortschritte bei der Segmentierung in Tomographiedaten.:Abstract IV Zusammenfassung VI 1 Introduction 1 1.1 Thesis Structure 2 1.2 Scientific Context 3 1.2.1 Developments in the Segmentation in Tomography Data 3 1.2.2 3D Semantic Segmentation using Machine Learning 5 1.2.3 Data Augmentation 6 2 Developed Software Solutions: AiSeg and unpAIred 9 2.1 Software Design 10 2.2 Installation 11 2.3 AiSeg 11 2.4 unpAIred 12 2.5 Limitations 12 3 Factors Affecting Image Quality in Computed Tomography 13 3.1 From CT Scan to Reconstruction 13 3.2 X-ray Tube and Focal Spot 14 3.3 Beam Hardening 14 3.4 Absorption, Scattering and Pairing 15 3.5 X-ray Detector 16 3.6 Geometric Calibration 17 3.7 Reconstruction Algorithm 17 3.8 Artifact corrections 18 4 On the Development of Augmentation Pipelines for Image Segmentation 19 4.0 Abstract 20 4.1 Introduction 20 4.2 Methods 21 4.2.1 Data Preparation 21 4.2.2 Augmentation 21 4.2.3 Networks 24 4.2.4 Training and Metrics 25 4.3 Experimental Design 26 4.3.1 Hardware 26 4.3.2 Workflow 26 4.3.3 Test on Cityscapes 26 4.4 Results and Discussion 26 4.4.1 Stage 1: Crating a Baseline 27 4.4.2 Stage 2: Using Offline Augmentation 27 4.4.3 Stage 3: Using Online Augmentation 27 4.4.4 Test on Cityscapes 29 4.4.5 Future Work – A New Online Augmentation 30 4.5 Conclusion 31 4.6 Appendix 31 4.6.1 Appendix A. List of All Networks 31 4.6.2 Appendix B. Augmentation Methods 32 4.6.3 Appendix C. Used RIWA Online Augmentation Parameters 36 4.6.4 Appendix D. Used Cityscapes Online Augmentation Parameters 36 4.6.5 Appendix E. Comparison of CNNs with best Backbones on RIWA 37 4.6.6 Appendix F. Segmentation Results 38 4.7 References 39 5 Comparison of 3D CNNs for Volume Segmentation 43 5.0 Abstract 44 5.1 Introduction 44 5.2 Datasets 44 5.2.1 Carbon Rovings 45 5.2.2 Concrete Pores 45 5.2.3 Polyethylene Fibers 45 5.2.4 Brain Mitochondria 45 5.2.5 Brain Tumor Segmentation Challenge (BraTS) 46 5.2.6 Head and Neck Cancer 46 5.3 Methods 46 5.3.1 Data Preprocessing 46 5.3.2 Hyperparameters 46 5.3.3 Metrics 47 5.3.4 Experimental Design 48 5.4 Results and Discussion 48 5.4.1 Impact of Initial Random States (Head and Neck Cancer Dataset) 48 5.4.2 Carbon Rovings 48 5.4.3 Concrete Pores 49 5.4.4 Polyethylene Fibers 49 5.4.5 Brain Mitochondria 50 5.4.6 BraTS 51 5.5 Conclusion 51 5.6 References 52 6 Segmentation of Carbon in CRC Using 3D Augmentation 55 6.0 Abstract 56 6.1 Introduction 56 6.2 Materials and Methods 58 6.2.1 Specimens 58 6.2.2 Microtomography 59 6.2.3 AI-Based Segmentation 60 6.2.4 Roving Extraction 64 6.2.5 Multiscale Modeling 65 6.2.6 Scaled Boundary Isogeometric Analysis 66 6.2.7 Parameterized RVE and Definition of Characteristic Geometric Properties 67 6.3 Results and Discussion 70 6.3.1 Microtomography 70 6.3.2 Deep Learning 71 6.3.3 Roving Extraction 74 6.3.4 Parameterized RVE and Definition of Characteristic Geometric Properties 75 6.4 Conclusion 79 6.5 References 80 7 Image-to-Image Translation for Semi-Supervised Semantic Segmentation 85 7.1 Introduction 85 7.2 Methods 86 7.2.1 Generative Adversarial Networks 87 7.2.2 Contrastive Unpaired Translation 87 7.2.3 Fréchet Inception Distance 89 7.2.4 Datasets 89 7.3 Experimental Design 92 7.4 Results and Discussion 94 7.4.1 Training and Inference of CUT 94 7.4.2 End-to-End Training for Semantic Segmentation 99 7.5 Conclusion 104 7.5.1 Future Work 104 8 Synthesis 107 8.1 Research Summary 107 8.1.1 Augmentation Pipelines 107 8.1.2 3D CNN Comparison 108 8.1.3 3D Data Augmentation for the Segmentation of Carbon Rovings 108 8.1.4 Synthetic Training Data Generation 109 8.2 Future Developments 109 8.2.1 Augmentation 109 8.2.2 Pre-trained 3D Encoder 111 8.2.3 On the Quality Control of Carbon Reinforced Concrete 111 8.2.4 Subvoxel Accurate Segmentation 113 8.2.5 Towards Volume-to-Volume Translation 114 8.3 Conclusion 114 References 117 List of Tables 125 List of Figures 127 List of Abbreviations 131

No description available.

Das vorliegende Heft 25/2022 der „Diskussionsbeiträge zur Kartosemiotik und zur Theorie der Kartographie“ enthält sechs Artikel und zwei Kurzberichte. Ausgabe 25 ist eine Jubiläumsausgabe, die sich dem Karten- und Atlasthema widmet. 1998 erschien der erste Sammelband „Diskussionsbeiträge“. Im Jahr 2022 wurde die fünfundzwanzigste Sammlung zur Veröffentlichung vorbereitet, die eine Reihe von Artikeln zu theoretischen Problemen der Kartographie und der Karto/Atlassemiotik fortsetzt. / The present issue 25 contains six articles and two reports. Issue 25 is an anniversary issue dedicated to the map and atlas theme. In 1998 the first anthology 'Discussion Papers on Cartosemiotics and Cartography Theory' was published. In 2022 the twenty-fifth collection was prepared for publication, continuing a series of articles on theoretical problems of cartography and carto/atlassemiotics. / Сборник «Дискуссионные статьи по картосемиотике и теории картографии» № 25, 2022 года содержит шесть статей и два сообщения. Сборник № 25 — это юбилейное издание, посвященное карто-атласной тематике. В 1998 году был издан первый сборник, в 2022 году был подготовлен к изданию двадцать пятый сборник, который продолжает серию отдельных статей по теоретическим проблемам картографии и карто-атласной семиотике.

Aerosols show type-specific characteristics, which depend on intensive aerosol optical and microphysical properties that influence the radiation processes in the atmosphere in several ways. There are still large uncertainties in the calculation of the aerosol direct radiative effect. The classification of aerosols and the characterization of the vertical aerosol distribution is needed in order to provide more accurate information for radiative-transfer simulations. In the framework of the present thesis, the vertical and spatial distribution as well as optical properties of atmospheric aerosols over the European continent were investigated based on lidar measurements. Possibilities for an aerosol classification or so-called aerosol typing were presented and major aerosol types were specified. Former studies about the classification of aerosols were summarized and representative values for aerosol-type-dependent parameters were given. Case studies were used to demonstrate how observations of the European lidar network EARLINET from 2008 until 2010 were analyzed for aerosol layers and how model simulations and auxiliary data including the assessment of meteorological conditions were applied to determine the origin of each single aerosol layer. Thus, aerosol-type dependent parameters were evaluated and a novel method for the typing of aerosols was developed, which can be used, e.g., within algorithms of satellite data retrievals. Additionally, conversion factors were determined, which are needed for the harmonization of satellite data of present and upcoming missions. Furthermore, findings of the aerosol typing based on EARLINET data were compared to results of the aerosol classification scheme for satellite-borne lidar measurements onboard CALIPSO. It could be shown that deficient classifications of the aerosol type emerged systematically within the automated CALIPSO algorithm. Those wrong classification leads to an underestimation of the single-scattering albedo and hence to an overestimation of the warming effect of the respective aerosol layer. This overestimated warming effect has to be kept in mind for simulations of the global aerosol radiative effect based on CALIPSO data. / Die Bestimmung des direkten Strahlungsantriebs von Aerosolen ist mit großen Unsicherheiten behaftet. Inwiefern Aerosole die Strahlungsprozesse in der Atmosphäre beeinflussen ist abhängig von ihren optischen und mikrophysikalischen Eigenschaften. Zur Optimierung von Strahlungstransfersimulationen werden daher ergänzende Informationen über typspezifische Aerosoleigenschaften sowie die vertikale Aerosolverteilung benötigt. Im Rahmen der vorliegenden Arbeit wurden anhand von Lidarmessungen die vertikale und räumliche Verteilung atmosphärischer Aerosole über Europa analysiert sowie deren optische Eigenschaften ermittelt. Einleitend werden Möglichkeiten der Aerosolklassifizierung erläutert und Aerosoltypen spezifiziert, die über Europa beobachtet werden können. Vorherige Studien zur Aerosolklassifizierung sind in einer Literaturübersicht zusammengefasst. Anhand von Fallstudien wurde zunächst die Analyse von Beobachtungen des europäischen Lidarnetzwerkes EARLINET von 2008 bis 2010 auf das Vorhandensein von Aerosolschichten verdeutlicht. Die Herkunft jeder einzelnen Aerosolschicht wurde anschließend unter Verwendung von Modellrechnungen sowie weiteren Informationen bestimmt und aerosoltypspezifische Kenngrößen berechnet. Mit Hilfe dieser Kenngrößen ist es möglich, den Typ des Aerosols abzuleiten. Daraus wurde eine neuartige Methode zur Typisierung von Aerosolen entwickelt, die z.B. in Algorithmen zur Verarbeitung von Satellitendaten verwendet werden kann. Zusätzlich wurden Umrechnungsfaktoren bestimmt, die zur Zusammenführung und zum Vergleich von Daten aktueller und zukünftiger Satellitenmissionen benötigt werden. Die Ergebnisse der Aerosoltypisierung auf Basis von EARLINET-Daten wurden anschließend mit Ergebnissen der automatischen Typisierung weltraumbasierter Lidarmessungen des CALIPSO-Satelliten verglichen. Es konnte gezeigt werden, dass innerhalb des CALIPSO-Algorithmus systematisch fehlerhafte Klassifizierungen des Aerosoltyps auftreten. Diese falsche Klassifizierung führt zu einer Unterschätzung der Einfachstreualbedo und zu einer Überschätzung der erwärmenden Wirkung der betreffenden Aerosolschicht. Die überschätzte Wärmewirkung hat wiederum fehlerhafte Ergebnisse bei Strahlungstransferrechnungen, die auf CALIPSO-Daten basieren, zur Folge.

Global climate models simulate a robust increase of global mean precipitation of about 1.5 to 2% per Kelvin surface warming in response to greenhouse gas (GHG) forcing.Here, it is shown that the sensitivity to aerosol cooling is robust as well, albeit roughly twice as large. This larger sensitivity is consistent with energy budget arguments. At the same time, it is still considerably lower than the 6.5 to 7% K−1 decrease of the water vapor concentration with cooling from anthropogenic aerosol because the water vapor radiative feedback lowers the hydrological sensitivity to anthropogenic forcings. When GHG and aerosol forcings are combined, the climate models with a realistic 20th century warming indicate that the globa lmean precipitation increase due to GHG warming has, until recently, been completely masked by aerosol drying. This explains the apparent lack of sensitivity of the global mean precipitation to the net global warming recently found in observations. As the importance of GHG warming increases in the future, a clear signal will emerge.

Die vorliegende Arbeit befasst sich mit dem Einfluss einer hyperthermischen Stickstoffionenbestrahlung (Ekin < 25 eV) auf das Galliumnitrid-Schichtwachstum. Dabei wird insbesondere der Einfluss einer Oberflächenrekonstruktion, einer Strukturierung der Oberfläche, einer Zwischenschicht (Pufferschicht) und der Einfluss verschiedener Siliziumsubstratorientierungen auf das epitaktische Wachstum von dünnen Galliumnitrid-Schichten nach einer hyperthermischen Stickstoffionenbestrahlung diskutiert. Ziel war es, möglichst dünne, epitaktische und defektarme Galliumnitrid-Schichten zu erhalten. Für die Charakterisierung der Galliumnitrid-Schichten und der Siliziumsubstrate standen diverse Analysemethoden zur Verfügung. Die kristalline Oberflächenstruktur konnte während des Wachstums mittels Reflexionsbeugung hochenergetischer Elektronen beobachtet werden. Nachfolgend wurde die Oberflächentopografie, die kristalline Struktur und Textur, sowie die optischen Eigenschaften der Galliumnitrid-Schichten mittels Rasterkraftmikroskopie, Röntgenstrahl-Diffraktometrie, hochauflösender Transmissionselektronenmikroskopie und Photolumineszenzspektroskopie untersucht.