Harz University of Applied Sciences

Accurate segmentation and analysis of root images from soil-grown plants are critical for advancing our understanding of root growth and plasticity under varying environmental conditions. Most approaches typically rely on binary segmentation of the entire root system architecture (RSA), which limits their ability to capture the hierarchical complexity of root structures, including axial and lateral roots. To address this, our study evaluated five convolutional neural network (CNN) architectures for multi-class (i.e. axial/lateral) segmentation of 2D root images from barley plants grown in rhizoboxes: (i) U-Net, (ii) U-Net with Atrous Spatial Pyramid Pooling (UnetASPP), (iii) U-Net with Attention Block (UnetAtt), (iv) DeepLabV3+ with MobileNetV2 (DLMB), and (v) DeepLabV3+ with ResNet-50 (DLR50). Among these, the DLR50 model achieved the highest segmentation accuracy, particularly for distinguishing lateral roots within complex RSA structures. Furthermore, analysis of root traits derived from the segmented images confirmed that DLR50 produced the most reliable estimations of phenotypic traits compared to ground truth measurements. These findings highlight the strong potential of advanced multi-class CNN models—especially DLR50—for detailed and quantitative analysis of soil-root systems, providing new insights into root responses to environmental conditions. • Evaluated five CNN models for multi-class segmentation of 2D root images, distinguishing axial and lateral roots. • DLR50 (DeepLabV3+ with ResNet-50) outperformed all models in accurately segmenting complex root structures. • Improved trait prediction accuracy using DLR50-segmented images compared to ground truth data. • Enables fine-grained analysis of root plasticity in response to environmental changes.

Software engineering is a knowledge-intensive process. Consequently, researchers typically understand any lack of knowledge as a problem that must be mitigated by improving, for instance, program comprehension, reverse engineering, community collaboration, or documentation. However, a lack of knowledge may not always be a problem. In fact, research in the field of ignorance studies has highlighted the diversity of ignorance phenomena and emphasized their relevance in social interaction. We argue that focusing on non-knowledge as one of these phenomena can also be useful for software-engineering research. In this paper, we develop and substantiate this claim by providing a brief overview on the (social scientific) research on ignorance and by proposing a working definition of non-knowledge for software-engineering research. Then, we sketch how the perspective of non-knowledge as a social phenomenon can benefit future research within software engineering and propose three concrete directions to investigate. We envision that non-knowledge contributes a new lens to manage the complexity of intellectual capital and knowledge in modern software engineering. With this paper, we hope to motivate and guide future software-engineering research in this direction.

Abstract A particularly controversial corporate governance practice is the case of former CEOs who decide - and are allowed - to extend their influence by remaining as chairs of the supervisory board (in this study referred to as CACs: CEOs as Chairs). We analyze the effects and preconditions of CACs and confirm a formerly observed pattern that departing CEOs who remain as board chairs restrict their successors’ potential to initiate changes. However, inhibited change is intended and will continue even after the CAC has finally left the scene, i.e., passing the baton or the ultimate departure of the CAC becomes actually a ‘non-event’. In a German context, this commitment to the status quo is essentially good news: By analyzing German HDAX firms over a period of twenty years, we find empirical support that it is mainly CEOs effectively meeting the expectations of two powerful stakeholder groups (namely, shareholders and employees) who get the chance to continue as board chairs and that this practice pays off for both stakeholder groups in the long run. Consequently, the installation of a CAC is not necessarily a symptom of a missed opportunity for strategic realignment, but can rather be an indicator of a firm’s sustainable development.

Weather radars are essential in the Quantitative Precipitation Estimates (QPE) but are susceptible to calibration errors. Previous work demonstrated that observations from the Ku-band Dual Polarization Radar (DPR) radar on board the Global Precipitation Measurement Mission Dual-Precipitation Radar (GPM) are suitable for ground radar calibration. Several studies volume-matched ground radar and GPM DPR Ku-band reflectivities for the absolute calibration of ground radars, by applying different constraints and filters in the volume-matching procedure. This study compares and evaluates volume-matching thresholds and data filtering schemes for the Rizoelia, Larnaca (LCA) and Nata, Pafos (PFO) radars of the Cyprus weather radar network from October 2017 till May 2023. Excluding reflectivities below and within the melting layer with a 250 m buffer yielded consistent results for both ground radars. The selected calibration schemes were combined, and the resulting offsets were compared to stable radar parameters to identify stable calibration periods. The consistency of the wet hydrological year October 2019 to September 2020 suggests that radar calibration results are prone to differences in meteorological conditions, as scarce rainfall can result in insufficient data for reliable calibration. Future work will incorporate disdrometer measurements and extend the analysis to quantitative precipitation estimation.

Abstract. A simulation study of the potential impact of wildfire smoke on Arctic cirrus formation is presented. The simulations complement the MOSAiC (Multidisciplinary drifting Observatory for the Study of Arctic Climate) field observations, discussed in Part 1 (Ansmann et al., 2025) of this work. The observations suggest that Siberian wildfire smoke had a strong impact on Arctic cirrus formation in the winter of 2019–2020. Via simulations, a detailed insight into the potential of wildfire smoke to influence Arctic cirrus formation as a function of observed meteorological and environmental conditions (temperature, relative humidity, large-scale and gravity-wave-induced lofting conditions, and ice-nucleating particle (INP) concentration) is provided. Lidar-derived values of the INP concentration serve as input, and ice crystal number concentration (ICNC) values retrieved from combined lidar–radar observations are used for comparison with the simulation results. The simulations show that the observed smoke pollution levels in the upper troposphere were high enough to trigger strong ice nucleation. The simulations also corroborate the hypothesis stated in Part 1 (Ansmann et al., 2025): the persistent smoke layer, continuously observed over the central Arctic during the winter half year 2019–2020, was able to widely suppress homogeneous freezing so that the smoke aerosol most probably controlled cirrus formation and properties. The observations suggest that the INP reservoir was continuously refilled from the lower stratosphere. Furthermore, the simulations confirm that the observed high ice saturation ratios of 1.3–1.5 over the North Pole region at cirrus tops (with top temperatures of −60 to −75 °C) point to inefficient INPs, as expected when wildfire smoke particles (organic particles) serve as INPs. Finally, the simulations revealed that ice nucleation in widespread and frequently occurring shallow updrafts (with low amplitudes) seems to be responsible for the observed low ICNC values of typically 1–50 crystals L−1 in the Arctic cirrus virga.

This article explores how local governments apply strategic management principles to implement the 2030 Agenda for Sustainable Development. The research is based on a multiple case study, drawing on document analysis and interviews conducted in three German municipalities. Findings reveal a spectrum of approaches, ranging from emergent, informal strategies to highly formalized frameworks with measurable targets and indicators. Although formalization is driven by external funding requirements and reporting obligations, it risks creating administrative efforts that undermine effectiveness.

<strong class="journal-contentHeaderColor">Abstract.</strong> Height-resolved monitoring of life cycles of mixed-phase clouds (MPCs) was performed in the free troposphere over the central Arctic during the MOSAiC (Multidisciplinary drifting Observatory for the Study of Arctic Climate) expedition from October 2019 to September 2020. The research icebreaker <em>Polarstern</em> served as a platform for state-of-the-art remote sensing of aerosols and clouds. The use of the recently introduced dual field-of-view polarization lidar technique in combination with the well-established lidar-radar retrieval technique provided, for the first time, a robust instrumental basis to monitor the evolution of the liquid and the ice phase of MPCs and the interplay between the two phases. We discuss two long-lasting Arctic MPC cases observed close to the North Pole. During the late summer MPC event, most likely three gravity waves strongly disturbed the cloud evolution. We documented this perturbation in detail in terms of liquid and ice-phase properties and the recovery of the strongly disturbed liquid phase within a few hours. For the first time, cloud statistics, covering all seasons of a year, are presented for liquid-bearing stratiform clouds in the central Arctic. The focus is on the optical and microphysical properties of the liquid phase which is of key importance for a long MPC lifetime. The observations confirmed that ice formation occurs predominantly via immersion freezing. We also found that activation of aerosol particles to form droplets is of great importance for the longevity of MPCs and that the free tropospheric reservoirs of cloud-condensation nuclei and ice-nucleating particles seem to be usually well-filled.

In the past decades, ground-based weather radars gained popularity for enhancing the understanding of precipitation systems, the accuracy of the Quantitative Precipitation Estimation (QPE) and for serving as input in numerical weather models. Nevertheless, they are prone to errors from various sources, including significant calibration errors. Previous research showed that the Ku-band precipitation radar aboard the Global Precipitation Measurement Mission Dual-Precipitation Radar (GPM DPR) is effective for calibrating ground-based radars. Several studies proposed the alignment of ground-based radar reflectivities with those from the GPM DPR to achieve their absolute calibration. This study performs the absolute calibration of the Rizoelia (LCA) and Nata (PFO) radars in Cyprus for approximately six years of observations (October 2017 to May 2023), assessing and comparing volume-matching thresholds and data filtering techniques. The results indicate that excluding reflectivities within the melting layer and adding a 250 m buffer consistently improved calibration for both radars. The selected calibration schemes were combined, and the resulting offsets were compared against stable radar parameters to identify stable calibration periods. Future work will include disdrometer data and expand the analysis to quantitative precipitation estimation.AcknowledgementsThe authors acknowledge the &amp;#8216;EXCELSIOR&amp;#8217;: ERATOSTHENES: E&amp;#935;cellence Research Centre for Earth Surveillance and Space-Based Monitoring of the Environment H2020 Widespread Teaming project (www.excelsior2020.eu). The &amp;#8216;EXCELSIOR&amp;#8217; project has received funding from the European Union&amp;#8217;s Horizon 2020 research and innovation programme under Grant Agreement No 857510, from the Government of the Republic of Cyprus through the Directorate General for the European Programmes, Coordination and Development and the Cyprus University of Technology.The authors also acknowledge the Department of Meteorology of the Republic of Cyprus for providing the X-band radar data.&amp;#160;

ABSTRACT Control schemes are often used to monitor variables , but not all process data fit this description, as some data may actually be attributive in nature. For this reason, considerable attention has recently been paid to control schemes designed for attributes. In particular, new control schemes based on the hypergeometric distribution, namely hypergeometric p and np schemes, have been proposed. However, these schemes are mostly Shewhart‐type control schemes, and they are often criticized due to their inferior performance in detecting small and medium shifts. To address this issue, we present the exponentially weighted moving average (EWMA) hypergeometric np scheme in this paper. Similar to the hypergeometric np scheme, the proposed scheme is more practically convenient than the hypergeometric p scheme since it works with integer values. Since computing the run length properties for an EWMA scheme that depends on discrete data is challenging, we also consider the “continuousify” technique in this paper. We compare the introduced scheme with the existing hypergeometric np control scheme and demonstrate that the former scheme outperforms the latter scheme for all shift sizes. Furthermore, we investigate the optimal design of the EWMA hypergeometric np scheme to enhance its practicality and illustrate its application on a real dataset.