Refine
Has Fulltext
- yes (165) (remove)
Year of publication
- 2021 (165) (remove)
Document Type
- Doctoral Thesis (165) (remove)
Is part of the Bibliography
- yes (165)
Keywords
- Spektroskopie (4)
- Klimawandel (3)
- Politik (3)
- climate change (3)
- spectroscopy (3)
- 3D-Visualisierung (2)
- Agrarökologie (2)
- Air pollution (2)
- Alpen (2)
- Alps (2)
Institute
- Institut für Physik und Astronomie (24)
- Institut für Geowissenschaften (22)
- Institut für Biochemie und Biologie (20)
- Institut für Chemie (20)
- Hasso-Plattner-Institut für Digital Engineering GmbH (13)
- Institut für Umweltwissenschaften und Geographie (9)
- Wirtschaftswissenschaften (9)
- Institut für Informatik und Computational Science (8)
- Department Psychologie (6)
- Extern (5)
- Institut für Mathematik (5)
- Institut für Ernährungswissenschaft (4)
- Department Erziehungswissenschaft (3)
- Department Sport- und Gesundheitswissenschaften (3)
- Historisches Institut (3)
- Department Linguistik (2)
- Institut für Romanistik (2)
- Department Grundschulpädagogik (1)
- Department Musik und Kunst (1)
- Department für Inklusionspädagogik (1)
- Fachgruppe Betriebswirtschaftslehre (1)
- Foundations of Computational Linguistics (1)
- Institut für Germanistik (1)
- Institut für Jüdische Studien und Religionswissenschaft (1)
- Institut für Philosophie (1)
- Potsdam Institute for Climate Impact Research (PIK) e. V. (1)
- Psycholinguistics and Neurolinguistics (1)
- Sozialwissenschaften (1)
- Strukturbereich Kognitionswissenschaften (1)
- Öffentliches Recht (1)
While the last few decades have seen impressive improvements in several areas in Natural Language Processing, asking a computer to make sense of the discourse of utterances in a text remains challenging. There are several different theories that aim to describe and analyse the coherent structure that a well-written text inhibits. These theories have varying degrees of applicability and feasibility for practical use. Presumably the most data-driven of these theories is the paradigm that comes with the Penn Discourse TreeBank, a corpus annotated for discourse relations containing over 1 million words. Any language other than English however, can be considered a low-resource language when it comes to discourse processing.
This dissertation is about shallow discourse parsing (discourse parsing following the paradigm of the Penn Discourse TreeBank) for German. The limited availability of annotated data for German means the potential of modern, deep-learning based methods relying on such data is also limited. This dissertation explores to what extent machine-learning and more recent deep-learning based methods can be combined with traditional, linguistic feature engineering to improve performance for the discourse parsing task. A pivotal role is played by connective lexicons that exhaustively list the discourse connectives of a particular language along with some of their core properties.
To facilitate training and evaluation of the methods proposed in this dissertation, an existing corpus (the Potsdam Commentary Corpus) has been extended and additional data has been annotated from scratch. The approach to end-to-end shallow discourse parsing for German adopts a pipeline architecture and either presents the first results or improves over state-of-the-art for German for the individual sub-tasks of the discourse parsing task, which are, in processing order, connective identification, argument extraction and sense classification. The end-to-end shallow discourse parser for German that has been developed for the purpose of this dissertation is open-source and available online.
In the course of writing this dissertation, work has been carried out on several connective lexicons in different languages. Due to their central role and demonstrated usefulness for the methods proposed in this dissertation, strategies are discussed for creating or further developing such lexicons for a particular language, as well as suggestions on how to further increase their usefulness for shallow discourse parsing.
The majority of baryons in the Universe is believed to reside in the intergalactic medium (IGM). This makes the IGM an important component in understanding cosmological structure formation. It is expected to trace the same dark matter distribution as galaxies, forming structures like filaments and clusters. However, whereas galaxies can be observed to be arranged along these large-scale structures, the spatial distribution of the diffuse IGM is not as easily unveiled. Absorption line studies of quasar (QSO) spectra can help with mapping the IGM, as well as the boundary layer between IGM and galaxies: the circumgalactic medium (CGM). By studying gas in the Local Group, as well as in the IGM, this study aims to get a better understanding of how the gas is linked to the large-scale structure of the local Universe and the galaxies residing in that structure.
Chapter 1 gives an introduction to the CGM and IGM, while the methods used in this study are explained in Chapter 2. Chapter 3 starts on a relatively small cosmological scale, namely that of our Local Group, which includes i.a. the Milky Way (MW) and the M31. Within the CGM of the MW, there exist denser clouds, some of which are infalling while others are moving away from the Galactic disc. To study these clouds, 29 QSO spectra obtained with the Cosmic Origins Spectrograph (COS) aboard the Hubble Space Telescope (HST) were analysed. Abundances of Si II, Si III, Si IV, C II, and C IV were measured for 69 HVCs belonging to two samples: one in the direction of the LG’s barycentre and the other in the anti-barycentre direction. Their velocities range from -100 ≥ vLSR ≥ -400 km/s for the barycentre sample and between +100 ≤ vLSR ≤ +300 km/s for the anti-barycentre sample. By using Cloudy models, these data could then be used to derive gas volume densities for the HVCs. Because of the relationship between density and pressure of the ambient medium, which is in turn determined by the Galactic radiation field, the distances of the HVCs could be estimated. From this, a subsample of absorbers located in the direction of M31 was found to exist outside of the MW’s virial radius, their low densities (log nH ≤ -3.54) making it likely for them to be part of the gas in between the MW and M31. No such low-density absorbers were found in the anti-barycentre sample. Our results thus hint at gas following the dark matter potential, which would be deeper between the MW and M31 as they are by far the most massive members of the LG.
From this bridge of gas in the LG, this study zooms out to the large-scale structure of the local Universe (z ~ 0) in Chapter 4. Galaxy data from the V8k catalogue and QSO spectra from COS were used to study the relation between the galaxies tracing large-scale filaments and the gas existing outside of those galaxies. This study used the filaments defined in Courtois et al. (2013). A total of 587 Lyman α (Lyα) absorbers were found in the 302 QSO spectra in the velocity range 1070 - 6700 km/s. After selecting sightlines passing through or close to these filaments, model spectra were made for 91 sightlines and 215 (227) Lyα absorbers (components) were measured in this sample. The velocity gradient along each filament was calculated and 74 absorbers were found within 1000 km/s of the nearest filament segment.
In order to find whether the absorbers are more tied to galaxies or to the large-scale structure, equivalent widths of the Lyα absorbers were plotted against both galaxy and filament impact parameters. While stronger absorbers do tend to be closer to either galaxies or filaments, there is a large scatter in this relation. Despite this large scatter, this study found that the absorbers do not follow a random distribution either. They cluster less strongly around filaments than galaxies, but stronger than random distributions, as confirmed by a Kolmogorov-Smirnov test.
Furthermore, the column density distribution function found in this study has a slope of -β = 1.63±0.12 for the total sample and -β =1.47±0.24 for the absorbers within 1000 km/s of a filament. The shallower slope for the latter subsample could indicate an excess of denser absorbers within the filament, but they are consistent within errors. These values are in agreement with values found in e.g. Lehner et al. (2007); Danforth et al. (2016).
The picture that emerges from this study regarding the relation between the IGM and the large-scale structure in the local Universe fits with what is found in other studies: while at least part of the gas traces the same filamentary structure as galaxies, the relation is complex. This study has shown that by taking a large sample of sightlines and comparing the data gathered from those with galaxy data, it is possible to study the gaseous large-scale structure. This approach can be used in the future together with simulations to get a better understanding of structure formation and evolution in the Universe.
In C3 plants, CO2 diffuses into the leaf and is assimilated by the Calvin-Benson cycle in the mesophyll cells. It leaves Rubisco open to its side reaction with O2, resulting in a wasteful cycle known as photorespiration. A sharp fall in atmospheric CO2 levels about 30 million years ago have further increased the side reaction with O2. The pressure to reduce photorespiration led, in over 60 plant genera, to the evolution of a CO2-concentrating mechanism called C4 photosynthesis; in this mode, CO2 is initially incorporated into 4-carbon organic acids, which diffuse to the bundle sheath and are decarboxylated to provide CO2 to Rubisco. Some genera, like Flaveria, contain several species that represent different steps in this complex evolutionary process. However, the majority of terrestrial plant species did not evolve a CO2-concentrating mechanism and perform C3 photosynthesis.
This thesis compares photosynthetic metabolism in several species with C3, C4 and intermediate modes of photosynthesis. Metabolite profiling and stable isotope labelling were performed to detect inter-specific differences changes in metabolite profile and, hence, how a pathway operates. The results obtained were subjected to integrative data analyses like hierarchical clustering and principal component analysis, and were deepened by correlation analyses to uncover specific metabolic features and reaction steps that were conserved or differed between species.
The main findings are that Calvin-Benson cycle metabolite profiles differ between C3 and C4 species and between different C3 species, including a very different response to rising irradiance in Arabidopsis and rice. These findings confirm Calvin-Benson cycle operation diverged between C3 and C4 species and, most unexpectedly, even between different C3 species. Moreover, primary metabolic profiles supported the current C4 evolutionary model in the genus Flaveria and also provided new insights and opened up new questions. Metabolite profiles also point toward a progressive adjustment of the Calvin-Benson cycle during the evolution of C4 photosynthesis. Overall, this thesis point out the importance of a metabolite-centric approach to uncover underlying differences in species apparently sharing the same photosynthetic routes and as a valid method to investigate evolutionary transition between C3 and C4 photosynthesis.
One third of the world's population lives in areas where earthquakes causing at least slight damage are frequently expected. Thus, the development and testing of global seismicity models is essential to improving seismic hazard estimates and earthquake-preparedness protocols for effective disaster-risk mitigation. Currently, the availability and quality of geodetic data along plate-boundary regions provides the opportunity to construct global models of plate motion and strain rate, which can be translated into global maps of forecasted seismicity. Moreover, the broad coverage of existing earthquake catalogs facilitates in present-day the calibration and testing of global seismicity models. As a result, modern global seismicity models can integrate two independent factors necessary for physics-based, long-term earthquake forecasting, namely interseismic crustal strain accumulation and sudden lithospheric stress release.
In this dissertation, I present the construction of and testing results for two global ensemble seismicity models, aimed at providing mean rates of shallow (0-70 km) earthquake activity for seismic hazard assessment. These models depend on the Subduction Megathrust Earthquake Rate Forecast (SMERF2), a stationary seismicity approach for subduction zones, based on the conservation of moment principle and the use of regional "geodesy-to-seismicity" parameters, such as corner magnitudes, seismogenic thicknesses and subduction dip angles. Specifically, this interface-earthquake model combines geodetic strain rates with instrumentally-recorded seismicity to compute long-term rates of seismic and geodetic moment. Based on this, I derive analytical solutions for seismic coupling and earthquake activity, which provide this earthquake model with the initial abilities to properly forecast interface seismicity. Then, I integrate SMERF2 interface-seismicity estimates with earthquake computations in non-subduction zones provided by the Seismic Hazard Inferred From Tectonics based on the second iteration of the Global Strain Rate Map seismicity approach to construct the global Tectonic Earthquake Activity Model (TEAM). Thus, TEAM is designed to reduce number, and potentially spatial, earthquake inconsistencies of its predecessor tectonic earthquake model during the 2015-2017 period. Also, I combine this new geodetic-based earthquake approach with a global smoothed-seismicity model to create the World Hybrid Earthquake Estimates based on Likelihood scores (WHEEL) model. This updated hybrid model serves as an alternative earthquake-rate approach to the Global Earthquake Activity Rate model for forecasting long-term rates of shallow seismicity everywhere on Earth.
Global seismicity models provide scientific hypotheses about when and where earthquakes may occur, and how big they might be. Nonetheless, the veracity of these hypotheses can only be either confirmed or rejected after prospective forecast evaluation. Therefore, I finally test the consistency and relative performance of these global seismicity models with independent observations recorded during the 2014-2019 pseudo-prospective evaluation period. As a result, hybrid earthquake models based on both geodesy and seismicity are the most informative seismicity models during the testing time frame, as they obtain higher information scores than their constituent model components. These results support the combination of interseismic strain measurements with earthquake-catalog data for improved seismicity modeling. However, further prospective evaluations are required to more accurately describe the capacities of these global ensemble seismicity models to forecast longer-term earthquake activity.
The life cycle of higher plants is based on recurring phases of growth and development based on repetitive sequences of cell division, cell expansion and cell differentiation. This dissertation deals with two projects, each of them investigating two different topics that are related to cell expansion. The first project is examining an Arabidopsis thaliana mutant exhibiting overall cell enlargement and the second project is analysing two naturally occurring floral morphs of Amsinckia spectabilis (Boraginaceae) differing (amongst others) in style length and anther heights due to differences in longitudinal cell elongation. The EMS-mutant eop1 was shown to exhibit a petal size increase of 26% caused by cell enlargement. Further phenotypes were detected, such as cotyledon size increase (based on larger cells) as well as increased carpel, sepal, leaf and pollen sizes. Plant height was shown to be increased and more highly branched trichomes explained the hairy eop1 phenotype. Fine mapping revealed the causal SNP to be a C to T transition at the last nucleotide of intron 7 of the INCURVATA11 (ICU11) gene, a 2-oxoglutarate /Fe(II)-dependant dioxygenase, and thus causing missplicing of the mRNA. Two T-DNA insertion lines (icu11-2 & icu11-4) confirmed ICU11 as causal gene by exhibiting increased petal size. A comparison of three icu11 alleles, which possessed different mutation-related changes, either overexpressing ICU11 or modified mRNAs, was the base for investigating the molecular mechanism that underlies the observed phenotype. Different approaches revealed contradictory results regarding ICU11 protein functionality in the icu11 mutants. A complementation assay proved the three mutants to be exchangeable and ICU11 overexpression in the wild-type led to an icu11-like phenotype, arguing for all three icu11 mutants to be GOF mutants. Contradicting this conclusion, the icu11-4 line could be rescued by a genomic ICU11 transgene. A model, based on the assumption that an overexpression of ICU11 is inhibiting the function of the protein, and thus causing the same effect as a LOF protein was proposed. Further, icu11-3 (eop1) mutants were shown to have an increased resistance towards paclobutrazol, a gibberellin (GA) inhibitor and an upregulation of AtGA20ox2, a main GA biosynthesis gene. Additionally, ICU11 subcellular localization was discovered to be cytoplasmic, supporting the assumption, that ICU11 affects GA biosynthesis and overall GA level, possibly explaining the observed (GA-overdose) phenotype.
The second project aimed to identify the genetic base of the S-locus in Amsinckia spectabilis, as the Amsinckia genus represents untypical characteristics for a heterostylous species, such as no obvious self-incompatibility (SI) and the repeated transition towards homostylous and fully selfing variants. The work was based on three Amsinckia spectabilis forms: a heterostylous form, consisting of two floral morphs with reciprocal positioning of sexual organs (S-morph: high anthers and a short style and L-morph: low anthers and a long style), and two homostylous forms, one large-flowered and partially selfing and the other small-flowered and fully selfing. The maintenance of the two floral morphs is genetically based on the S-locus region, containing genes that encode for the morph-specific traits, which are marked by a tight linkage due to suppressed recombination. Natural populations are found to possess a 1:1 S:L morph ratio, that can be explained by predominant disassortative mating of the two morphs, causing the occurrence of the dominant S-allele only in the heterozygous state (heterozygous (Ss) for the S-morph and homozygous recessive (ss) for the L-morph). Investigation of morph-specific phenotypes detected 56% elongated L-morph styles and 58% higher positioned S-morph anthers. Approximately 50% of the observed size differences were explained by an increase in cell elongation. Moreover, additional phenotypes were found, such as 21% enlarged S-morph pollen and no obvious SI, confirmed by hand pollinated seed counts, in vivo pollen tube growth and the development of homozygous dominant SS individuals via selfing. The Amsinckia spec. S-locus was assumed to at least consist of the G- (style length), the A- (anther height) and the P- (pollen size) locus. Comparative Transcriptomics of the two morphs revealed 22 differentially expressed markers that were found to be located within two contigs of a SS individual PacBio genome assembly, allowing the localization of the S-locus to be delimited to a region of approximately 23 Mb. Contradictory to revealed S-loci within the plant kingdom, no strong argument for a present hemizygous region was found to be causal for the suppressed recombination of the S-locus, so that an inversion was assumed to be the causal mechanism.
Cyanobacteria are an abundant bacterial group and are found in a variety of ecological niches all around the globe. They can serve as a real threat for fish or mammals and can restrict the use of lakes or rivers for recreational purposes or as a source of drinking water, when they form blooms. One of the most abundant bloom-forming cyanobacteria is Microcystis aeruginosa.
In the first part of the study, the role and possible dynamics of RubisCO in M. aeruginosa during high-light irradiation were examined. Its response was analyzed on the protein and peptide level via immunoblotting, immunofluorescence microscopy and with high performance liquid chromatography (HPLC). It was revealed that large amounts of RubisCO were located outside of carboxysomes under the applied high light stress. RubisCO aggregated mainly underneath the cytoplasmic membrane. There it forms a putative Calvin-Benson-Bassham (CBB) super complex together with other enzymes of photosynthesis. This complex could be part of an alternative carbon-concentrating mechanism (CCM) in M. aeruginosa, which enables a faster, and energy saving adaptation to high light stress of the whole bloom.
Furthermore, the re-localization of RubisCO was delayed in the microcystin-deficient mutant ΔmcyB and RubisCO was more evenly distributed over the cell in comparison to the wild type. Since ΔmcyB is not harmed in its growth, possibly other produced cyanopeptides as aeruginosin or cyanopeptolin also play a role in the stabilization of RubisCO and the putative CBB complex, especially in the microcystin-free mutant.
In the second part of this work, the possible role of microcystin as an extracellular signaling peptide during the diurnal cycle was studied. HPLC analysis showed a strong increase of extracellular microcystin in the wild type when the population entered nighttime and it resumed into the next day as well. Together with the increase of extracellular microcystin, a strong decrease of protein-bound intracellular microcystin was observed via immunoblot analysis. Interestingly, the signal of the large subunit of RubisCO (RbcL) also diminished when high amounts of microcystin were present in the surrounding medium. Microcystin addition experiments to M. aeruginosa WT and ΔmcyB cultures support this observation, since the immunoblot signal of both subunits of RubisCO and CcmK, a shell protein of carboxysomes, diminished after the addition of microcystin. In addition, the fluctuation of cyanopeptolin during the diurnal cycle indicates a more prominent role of other cyanopeptides besides microcystin as a signaling peptide, intracellularly as well as extracellularly.
Zentrales Element dieser Arbeit ist die Synthese und Charakterisierung praktisch nutzbarer Ionogele. Die Basis der Polymerionogele bildet das Modellpolymer Polymethylmethacrylat. Als Additive kommen ionische Flüssigkeiten zum Einsatz, deren Grundlage Derivate des vielfach verwendeten Imidazoliumkations sind. Die Eigenschaften der eingebetteten ionischen Flüssigkeiten sind für die Ionogele funktionsgebend. Die Funktionalität der jeweiligen Gele und damit der Transfer der Eigenschaften von ionischen Flüssigkeiten auf die Ionogele wurde in der vorliegenden Arbeit mittels zahlreicher Charakterisierungstechniken überprüft und bestätigt. In dieser Arbeit wurden durch Ionogelbildung makroskopische Ionogelobjekte in Form von Folien und Vliesen erzeugt. Dabei kamen das Filmgießen und das Elektrospinnen als Methoden zur Erzeugung dieser Folien und Vliese zum Einsatz, woraus jeweils ein Modellsystem resultiert. Dadurch wird die vorliegende Arbeit in die Themenkomplexe „elektrisch halbleitende Ionogelfolien“ und „antimikrobiell aktive Ionogelvliese“ gegliedert. Der Einsatz von triiodidhaltigen ionischen Flüssigkeiten und einer Polymermatrix in einem diskontinuierlichen Gießprozess resultiert in elektrisch halbleitenden Ionogelfolien. Die flexiblen und transparenten Folien können Mittelpunkt zahlreicher neuer Anwendungsfelder im Bereich flexibler Elektronik sein. Das Elektrospinnen von Polymethylmethacrylat mit einer ionischen Flüssigkeit führte zu einem homogen Ionogelvlies, welches ein Modell für die Übertragung antimikrobiell aktiver Eigenschaften ionischer Flüssigkeiten auf poröse Strukturen zur Filtration darstellt. Gleichzeitig ist es das erste Beispiel für ein kupferchloridhaltiges Ionogel. Ionogele sind attraktive Materialien mit zahlreichen Anwendungsmöglichkeiten. Mit der vorliegenden Arbeit wird das Spektrum der Ionogele um ein elektrisch halbleitendes und ein antimikrobiell aktives Ionogel erweitert. Gleichzeitig wurden durch diese Arbeit der Gruppe der ionischen Flüssigkeiten drei Beispiele für elektrisch halbleitende ionische Flüssigkeiten sowie zahlreiche kupfer(II)chloridbasierte ionische Flüssigkeiten hinzugefügt.
Precipitation forecasting has an important place in everyday life – during the day we may have tens of small talks discussing the likelihood that it will rain this evening or weekend. Should you take an umbrella for a walk? Or should you invite your friends for a barbecue? It will certainly depend on what your weather application shows.
While for years people were guided by the precipitation forecasts issued for a particular region or city several times a day, the widespread availability of weather radars allowed us to obtain forecasts at much higher spatiotemporal resolution of minutes in time and hundreds of meters in space. Hence, radar-based precipitation nowcasting, that is, very-short-range forecasting (typically up to 1–3 h), has become an essential technique, also in various professional application contexts, e.g., early warning, sewage control, or agriculture.
There are two major components comprising a system for precipitation nowcasting: radar-based precipitation estimates, and models to extrapolate that precipitation to the imminent future. While acknowledging the fundamental importance of radar-based precipitation retrieval for precipitation nowcasts, this thesis focuses only on the model development: the establishment of open and competitive benchmark models, the investigation of the potential of deep learning, and the development of procedures for nowcast errors diagnosis and isolation that can guide model development.
The present landscape of computational models for precipitation nowcasting still struggles with the availability of open software implementations that could serve as benchmarks for measuring progress. Focusing on this gap, we have developed and extensively benchmarked a stack of models based on different optical flow algorithms for the tracking step and a set of parsimonious extrapolation procedures based on image warping and advection. We demonstrate that these models provide skillful predictions comparable with or even superior to state-of-the-art operational software. We distribute the corresponding set of models as a software library, rainymotion, which is written in the Python programming language and openly available at GitHub (https://github.com/hydrogo/rainymotion). That way, the library acts as a tool for providing fast, open, and transparent solutions that could serve as a benchmark for further model development and hypothesis testing.
One of the promising directions for model development is to challenge the potential of deep learning – a subfield of machine learning that refers to artificial neural networks with deep architectures, which may consist of many computational layers. Deep learning showed promising results in many fields of computer science, such as image and speech recognition, or natural language processing, where it started to dramatically outperform reference methods.
The high benefit of using "big data" for training is among the main reasons for that. Hence, the emerging interest in deep learning in atmospheric sciences is also caused and concerted with the increasing availability of data – both observational and model-based. The large archives of weather radar data provide a solid basis for investigation of deep learning potential in precipitation nowcasting: one year of national 5-min composites for Germany comprises around 85 billion data points.
To this aim, we present RainNet, a deep convolutional neural network for radar-based precipitation nowcasting. RainNet was trained to predict continuous precipitation intensities at a lead time of 5 min, using several years of quality-controlled weather radar composites provided by the German Weather Service (DWD). That data set covers Germany with a spatial domain of 900 km x 900 km and has a resolution of 1 km in space and 5 min in time. Independent verification experiments were carried out on 11 summer precipitation events from 2016 to 2017. In these experiments, RainNet was applied recursively in order to achieve lead times of up to 1 h. In the verification experiments, trivial Eulerian persistence and a conventional model based on optical flow served as benchmarks. The latter is available in the previously developed rainymotion library.
RainNet significantly outperformed the benchmark models at all lead times up to 60 min for the routine verification metrics mean absolute error (MAE) and critical success index (CSI) at intensity thresholds of 0.125, 1, and 5 mm/h. However, rainymotion turned out to be superior in predicting the exceedance of higher intensity thresholds (here 10 and 15 mm/h). The limited ability of RainNet to predict high rainfall intensities is an undesirable property which we attribute to a high level of spatial smoothing introduced by the model. At a lead time of 5 min, an analysis of power spectral density confirmed a significant loss of spectral power at length scales of 16 km and below.
Obviously, RainNet had learned an optimal level of smoothing to produce a nowcast at 5 min lead time. In that sense, the loss of spectral power at small scales is informative, too, as it reflects the limits of predictability as a function of spatial scale. Beyond the lead time of 5 min, however, the increasing level of smoothing is a mere artifact – an analogue to numerical diffusion – that is not a property of RainNet itself but of its recursive application. In the context of early warning, the smoothing is particularly unfavorable since pronounced features of intense precipitation tend to get lost over longer lead times. Hence, we propose several options to address this issue in prospective research on model development for precipitation nowcasting, including an adjustment of the loss function for model training, model training for longer lead times, and the prediction of threshold exceedance.
The model development together with the verification experiments for both conventional and deep learning model predictions also revealed the need to better understand the source of forecast errors. Understanding the dominant sources of error in specific situations should help in guiding further model improvement. The total error of a precipitation nowcast consists of an error in the predicted location of a precipitation feature and an error in the change of precipitation intensity over lead time. So far, verification measures did not allow to isolate the location error, making it difficult to specifically improve nowcast models with regard to location prediction.
To fill this gap, we introduced a framework to directly quantify the location error. To that end, we detect and track scale-invariant precipitation features (corners) in radar images. We then consider these observed tracks as the true reference in order to evaluate the performance (or, inversely, the error) of any model that aims to predict the future location of a precipitation feature. Hence, the location error of a forecast at any lead time ahead of the forecast time corresponds to the Euclidean distance between the observed and the predicted feature location at the corresponding lead time.
Based on this framework, we carried out a benchmarking case study using one year worth of weather radar composites of the DWD. We evaluated the performance of four extrapolation models, two of which are based on the linear extrapolation of corner motion; and the remaining two are based on the Dense Inverse Search (DIS) method: motion vectors obtained from DIS are used to predict feature locations by linear and Semi-Lagrangian extrapolation.
For all competing models, the mean location error exceeds a distance of 5 km after 60 min, and 10 km after 110 min. At least 25% of all forecasts exceed an error of 5 km after 50 min, and of 10 km after 90 min. Even for the best models in our experiment, at least 5 percent of the forecasts will have a location error of more than 10 km after 45 min. When we relate such errors to application scenarios that are typically suggested for precipitation nowcasting, e.g., early warning, it becomes obvious that location errors matter: the order of magnitude of these errors is about the same as the typical extent of a convective cell. Hence, the uncertainty of precipitation nowcasts at such length scales – just as a result of locational errors – can be substantial already at lead times of less than 1 h. Being able to quantify the location error should hence guide any model development that is targeted towards its minimization. To that aim, we also consider the high potential of using deep learning architectures specific to the assimilation of sequential (track) data.
Last but not least, the thesis demonstrates the benefits of a general movement towards open science for model development in the field of precipitation nowcasting. All the presented models and frameworks are distributed as open repositories, thus enhancing transparency and reproducibility of the methodological approach. Furthermore, they are readily available to be used for further research studies, as well as for practical applications.
Mental health problems are highly prevalent worldwide. Fortunately, psychotherapy has proven highly effective in the treatment of a number of mental health issues, such as depression and anxiety disorders. In contrast, psychotherapy training as is practised currently cannot be considered evidence-based. Thus, there is much room for improvement. The integration of simulated patients (SPs) into psychotherapy training and research is on the rise. SPs originate from the medical education and have, in a number of studies, been demonstrated to contribute to effective learning environments. Nevertheless, there has been voiced criticism regarding the authenticity of SP portrayals, but few studies have examined this to date.
Based on these considerations, this dissertation explores SPs’ authenticity while portraying a mental disorder, depression. Altogether, the present cumulative dissertation consists of three empirical papers. At the time of printing, Paper I and Paper III have been accepted for publication, and Paper II is under review after a minor revision.
First, Paper I develops and validates an observer-based rating-scale to assess SP authenticity in psychotherapeutic contexts. Based on the preliminary findings, it can be concluded that the Authenticity of Patient Demonstrations scale is a reliable and valid tool that can be used for recruiting, training, and evaluating the authenticity of SPs.
Second, Paper II tests whether student SPs are perceived as more authentic after they receive an in-depth role-script compared to those SPs who only receive basic information on the patient case. To test this assumption, a randomised controlled study design was implemented and the hypothesis could be confirmed. As a consequence, when engaging SPs, an in-depth role-script with details, e.g. on nonverbal behaviour and feelings of the patient, should be provided.
Third, Paper III demonstrates that psychotherapy trainees cannot distinguish between trained SPs and real patients and therefore suggests that, with proper training, SPs are a promising training method for psychotherapy.
Altogether, the dissertation shows that SPs can be trained to portray a depressive patient authentically and thus delivers promising evidence for the further dissemination of SPs.
Polymeric semiconductors are strong contenders for replacing traditional inorganic semiconductors in electronic applications requiring low power, low cost and flexibility, such as biosensors, flexible solar cells and electronic displays. Molecular doping has the potential to enable this revolution by improving the conductivity and charge transport properties of this class of materials. Despite decades of research in this field, gaps in our understanding of the nature of dopant–polymer interactions has resulted in limited commercialization of this technology. This work aims at providing a deeper insight into the underlying mechanisms of molecular p-doping of semiconducting polymers in the solution and solid-state, and thereby bring the scientific community closer to realizing the dream of making organic semiconductors commonplace in the electronics industry. The role of 1) dopant size/shape, 2) polymer chain aggregation and 3) charge delocalization on the doping mechanism and efficiency is addressed using optical (UV-Vis-NIR) and electron paramagnetic resonance (EPR) spectroscopies. By conducting a comprehensive study of the nature and concentration of the doping-induced species in solutions of the polymer poly(3-hexylthiophene) (P3HT) with 3 different dopants, we identify the unique optical signatures of the delocalized polaron, localized polaron and charge-transfer complex, and report their extinction coefficient values. Furthermore, with X-ray diffraction, atomic force microscopy and electrical conductivity measurements, we study the impact of processing technique and doping mechanism on the morphology and thereby, charge transport through the doped films.
This work demonstrates that the doping mechanism and type of doping-induced species formed are strongly influenced by the polymer backbone arrangement rather than dopant shape/size. The ability of the polymer chain to aggregate is found to be crucial for efficient charge transfer (ionization) and polaron delocalization. At the same time, our results suggest that the high ionization efficiency of a dopant–polymer system in solution may subsequently hinder efficient charge transport in the solid-state due to the reduction in the fraction of tie chains, which enable charges to move efficiently between aggregated domains in the films. This study demonstrates the complex multifaceted nature of polymer doping while providing important hints for the future design of dopant-host systems and film fabrication techniques.