Search Results (2,098)

Pacific youth in Aotearoa New Zealand are culturally diverse and deeply rooted in their families and communities. Despite facing socioeconomic inequities, systemic barriers, and limited decision-making opportunities, they maintain a positive perception of health and actively contribute to collective wellbeing. This paper explores the strengths of Pacific youth and how these can be harnessed to mobilise community wellbeing and transformative change. Using Pacific research methodologies—lalaga (weaving) and talanoa—we integrate findings from three key sources: the Talavou o le Moana Pacific Youth19 Report (quantitative data from 1130 Pacific youth), the Pacific Youth Home and Family Brief (open-text responses on family life), and insights from a panel of Pacific policy, research, and community experts presented in a webinar. These resources were reviewed and woven together by a team of three Pacific practitioners and one New Zealand European researcher, all with backgrounds in youth health, social work, and Pacific education. The lalaga reveals Pacific youth’s collective strength, cultural identity, and deep sense of responsibility. Their resilience and leadership, even amid adversity, highlight the urgent need for culturally grounded, youth-led, and community-responsive approaches. Empowering Pacific youth as agents of change is essential for fostering holistic wellbeing and transformative futures. Full article

Conventional acoustic positioning systems are typically confined to regions where direct-path measurements are available. However, in long-range underwater environments, acoustic rays undergo multiple reflections at the sea surface and seafloor, complicating the modeling of sound speed and introducing uncertainty due to seafloor bathymetric errors. To address these challenges, a high-precision positioning technology suitable for long-range deep-sea scenarios is proposed. This technology constructs an effective sound speed model based on ray-tracing principles to accommodate multipath propagation. To mitigate model errors caused by inaccurate seafloor bathymetry, a sound speed compensation mechanism is introduced to enhance the precision of reflected-path measurements. The experimental results demonstrate that, with an array baseline of 8 km, the proposed method reduces the maximum ranging error over a 50 km horizontal distance from 137.9 m to 15.5 m. The root-mean-square positioning error is decreased from 157.9 m to 31.0 m, representing an improvement in positioning precision of 80.4%. These results confirm the feasibility of high-precision long-range acoustic positioning. Full article

The accelerating pace of urbanization has significantly complicated traffic management systems, leading to mounting challenges, such as persistent congestion, increased travel delays, and heightened environmental impacts. In response to these challenges, this study presents a novel deep learning framework designed to enhance short-term traffic flow prediction and support intelligent transportation systems within the context of smart cities. The proposed model integrates Gated Recurrent Units (GRUs) and Long Short-Term Memory (LSTM) networks, augmented by an attention mechanism that dynamically emphasizes relevant temporal patterns. The model was rigorously evaluated using the publicly available datasets and demonstrated substantial improvements over current state-of-the-art methods. Specifically, the proposed framework achieves a 3.75% reduction in the Mean Absolute Error (MAE), a 2.00% reduction in the Root Mean Squared Error (RMSE), and a 4.17% reduction in the Mean Absolute Percentage Error (MAPE) compared to the baseline models. The enhanced predictive accuracy and computational efficiency offer significant benefits for intelligent traffic control, dynamic route planning, and proactive congestion management, thereby contributing to the development of more sustainable and efficient urban mobility systems. Full article

Deep learning models have achieved significant progress in developing self-driving algorithms. Despite their advantages, these algorithms typically require substantial amounts of data for effective training. Critical driving data, in particular, is essential for enhancing training efficiency and ensuring driving safety. However, existing methods for identifying critical data often rely on human prior knowledge or are disconnected from the training of self-driving algorithms. In this paper, we introduce a novel data distillation technique designed to autonomously identify critical data for training self-driving algorithms. We conducted experiments with both numerical simulations and the NGSIM dataset, which consists of real-world car trajectories on highway US-101, to validate our approach. In the numerical experiments, the distillation method achieved a test root mean squared error of 1.933 using only 200 distilled training data samples, demonstrating a significant improvement in data efficiency compared to the 1.872 test error obtained with 20,000 randomly sampled training samples. The distilled critical data represents only 1% of the original dataset, optimizing data usage and significantly enhancing computational efficiency. For real-world NGSIM data, we demonstrate the performance of the proposed method in scenarios with extremely sparse data availability and show that our proposed data distillation method outperforms other sampling baselines, including Herding and K-centering. These experimental results highlight the capability of the proposed method to autonomously identify critical data without relying on human prior knowledge. Full article

Quantifying deep soil (10–100 cm) and non-growing-season soil respiration (SR) is crucial for refining carbon (C) cycle models, yet the regulatory mechanisms governing these processes remain unclear. The novelty of this study lies in its focus on deep soils and non-growing seasons to elucidate how soil properties regulate SR under these special conditions. We conducted an on-site field experiment in the Caragana korshinskii plantation, measuring SR at soil depths of 0–10 cm, 10–50 cm, and 50–100 cm during the non-growing season and growing. The results suggested that the annual cumulative soil CO₂ fluxes reached 510.1 (0–10 cm), 131.5 (10–50 cm), and 45.3 g CO₂·m⁻² (50–100 cm). These emissions during the non-growing season accounted for 33%, 31%, and 32%, respectively. The soil physical properties (temperature, moisture, bulk density) explained the greatest variation in SR during growing and non-growing periods, followed by the biological properties (α-diversity, root biomass) and chemical properties (soil organic C, ammonium nitrogen, total C/nitrogen ratio). Depth-specific analysis demonstrated that soil physical properties explained the most SR variance at three depths with independent contributions of 78.9% (0–10 cm), 89.7% (10–50 cm), and 76.9% (50–100 cm). These values exceeded the independent contributions of chemical properties (70.3%, 70.9%, 60.0%) and biological properties (54.9%, 45.1%, 41.6%) at the corresponding depths. Overall, deep soil and non-growing season SR represent important C emission sources; excluding them may therefore substantially overestimate net C sequestration potential. Full article

Due to its benefits and efficiency, mechanized sugarcane harvest is a common practice in Brazil; however, continuous traffic of agricultural machinery leads to soil compaction at the end of each harvest cycle. Hence, this study evaluated whether machine traffic affects soil physical and hydraulic properties, root growth, and crop productivity in sugarcane areas during different harvest cycles. Four treatments were performed consisting of an area planted with different stages (years) of sugarcane crop: T1 = after the first harvest—plant cane (area 1); T2 = after the second harvest—first ratoon cane (area 2); T3 = after the third harvest—second ratoon cane (area 3); T4 = after fourth harvest—third ratoon cane (area 4). Five sampling sites were considered in each area, constituting five replicates collected from four layers. Two collection positions were considered: wheel track (WT) and planting row (PR). Soil physical properties, root system, productivity, and biometric characteristics of the sugarcane crop were evaluated at depths of 0.00–0.05 m, 0.05–0.10 m, 0.10–0.20 m, and 0.20–0.40 m. Traffic during the sugarcane crop growth cycles affected soil physical and hydraulic properties, showing sensitivity to the effects of the different treatments, producing variations in root growth and crop productivity. Plant cane cycle showed lower soil penetration resistance, bulk density, microporosity, higher saturated soil hydraulic conductivity, and macroporosity when compared with the other cycles studied. In the 0.10–0.20 m layer, all treatments produced higher soil penetration resistance and density, and lower saturated soil hydraulic conductivity. Dry biomass, volume, and root area were higher for the plant cane cycle in the 0.00–0.05 m and 0.05–0.10 m layers compared with the other crop cycles. Root dry biomass is directly related to crop productivity in layers up to 0.40 m deep. Sugarcane productivity was affected along the crop cycles, with higher productivity observed in the plant cane and first ratoon cane cycles compared with the second and third ratoon cane cycles. Full article

This paper investigates the application of image segmentation techniques in endodontics, focusing on improving diagnostic accuracy and achieving faster segmentation by delineating specific dental regions such as teeth and root canals. Deep learning architectures, notably 3D U-Net and GANs, have advanced the image segmentation process for dental structures, supporting more precise dental procedures. However, challenges like the demand for extensive labeled datasets and ensuring model generalizability remain. Two semi-automatic segmentation workflows, Grow From Seeds (GFS) and Watershed (WS), were developed to provide quicker acquisition of ground truth training data for deep learning models using 3D Slicer software version 5.8.1. These workflows were evaluated against a manual segmentation benchmark and a recent dental segmentation automated tool on three separate datasets. The evaluations were performed by the overall shapes of a maxillary central incisor and a maxillary second molar and by the region of the root canal of both teeth. Results from Kruskal–Wallis and Nemenyi tests indicated that the semi-automated workflows, more often than not, were not statistically different from the manual benchmark based on dice coefficient similarity, while the automated method consistently provided significantly different 3D models from their manual counterparts. The study also explores the benefits of labor reduction and time savings achieved by the semi-automated methods. Full article

The scope of this research was to quantify the mid-term impact of different soil tillage on carbon/nitrogen agronomical key context under optimal growing conditions of the European moderate continental climate. A large-scale on-farm experiment was established in winter wheat/soybean two-crop long-term cultivation without fertilization on fertile Luvic Chernozem. Four treatments were conducted: (T1) ‘Deep Loosening’ with tillage depth of 50 cm, (T2) ‘Plowing’ to 30 cm, (T3) ‘Strip-Till’ with tillage depth of 20 cm, and (T4) ‘No-Till’; the tillage frequency at T1 and T2 was reduced and applied to soybean only, therefore, once per 2 years during the trial period 2020/21–2024/25. Unlike the crop yield, which decreased with tillage intensity decreasing (21.38 > 19.30 > 18.88 > 18.62 t/ha in dry matter cumulatively; T2 > T3 > T1 > T4), the carbon/nitrogen key agronomical parameters either increased (root nodules count/weight: thus confirmed convergent, occasionally reverse indicators; soil compaction: penetrometric resistance) or differed in varying patterns and extent (soil chemical indicators). In fertile Chernozem soils, tillage and indicators have different importance within the nexus studied; plowing still gives the most stable yields. To improve nitrogen fixing, farmers’ practices need to balance yield vs. soil health, including eliminating soil compaction. Full article

In lithium-ion batteries, the fracture of active particles that are under stress is a key cause of battery aging, which leads to a reduction in active materials, an increase in internal resistance, and a decay in battery capacity. A coupled electrochemical–thermal–mechanical model was established to study the concentration and stress distributions of negative electrode particles under different charging rates and ambient temperatures. The results show that during charging, the maximum lithium-ion concentration occurs on the particle surface, while the minimum concentration appears at the particle center. Moreover, as the temperature decreases, the concentration distribution of negative electrode active particles becomes more uneven. Stress analysis indicates that when charging at a rate of 1C and 0 °C, the maximum stress of particles at the negative electrode–separator interface reaches 123.7 MPa, while when charging at 30 °C, the maximum particle stress is 24.3 MPa. The maximum shear stress occurs at the particle center, presenting a tensile stress state, while the minimum shear stress is located on the particle surface, showing a compressive stress state. Finally, to manage the stress of active materials in lithium-ion batteries while charging for health maintenance, this study uses a DNN (Deep Neural Network) to predict the maximum shear stress of particles based on simulation results. The predicted indicators, MAE (Mean Absolute Error) and RMSE (Root Mean Square Error), are 0.034 and 0.046, respectively. This research is helpful for optimizing charging strategies based on the stress of active materials in lithium-ion batteries during charging, inhibiting battery aging and improving safety performance. Full article

To solve carrot grading problems (low manual efficiency, unquantifiable defects/secondary damage in machinery, gaps in slender carrot (aspect ratio > 4:1) sorting), this study develops a deep learning-based system. Methods: Build CarrotDSTNet (YOLOv8-seg + DeepSORT, optimized via DualConv/SegNeXt) for quality detection; adopt fuzzy comprehensive evaluation for grading; propose CarrotDTNet with an electronic fence for sorting. Results: Detection metrics improved; grading accuracy 94% (0.37 ms); sorting accuracy 97.39%, efficiency 310 roots/min. Contribution: Realizes non-contact, high precision/efficiency sorting, solves traditional issues, and supports carrot industry automation. Full article

This study investigated the Yongning Fortress ruins in Taiyuan through a comprehensive analytical approach employing scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), laser particle size analysis, X-ray diffraction (XRD), X-ray fluorescence spectroscopy (XRF), and ion chromatography (IC). The research focused on elucidating the disturbance mechanisms and environmental impacts induced by the root systems of five representative herbaceous species on rammed earth structures. The results demonstrated distinct, species-specific disturbance patterns. Melica roots created three-dimensional network damage, Artemisia capillaris primarily caused deep root penetration, Fallopia aubertii exhibited coupled physical–chemical effects, Convolvulus arvensis induced shallow horizontal expansion damage, while Cirsium formed a heterogeneous structure characterized by dense taproots and loose lateral roots. Environmental conditions, particularly moisture content, significantly influenced disturbance intensity. All root activities led to common deterioration processes, including particle rounding, gradation degradation, and formation of organic–mineral composites. Notably, vegetation markedly altered soluble salt distribution patterns, with Cirsium increasing total salt content to 3.7 times that of undisturbed rammed earth (0.48%), while sulfate ion concentration (1.16 × 10⁻³) approached hazardous thresholds. The study established a theoretical framework linking plant traits, disturbance mechanisms, and environmental response, and proposed risk-based zoning strategies for preservation. These outcomes provide significant theoretical foundations and practical guidance for the scientific conservation of rammed earth heritage sites. Full article

To address the challenges of multi-scale missed detections, false positives, and incomplete boundary segmentation in cotton field residual plastic film detection, this study proposes the RSE-YOLO-Seg model. First, a PKI module (adaptive receptive field) is integrated into the C3K2 block and combined with the SegNext attention mechanism (multi-scale convolutional kernels) to capture multi-scale residual film features. Second, RFCAConv replaces standard convolutional layers to differentially process regions and receptive fields of different sizes, and an Efficient-Head is designed to reduce parameters. Finally, an NM-IoU loss function is proposed to enhance small residual film detection and boundary segmentation. Experiments on a self-constructed dataset show that RSE-YOLO-Seg improves the object detection average precision (mAP50(B)) by 3% and mask segmentation average precision (mAP50(M)) by 2.7% compared with the baseline, with all module improvements being statistically significant (p < 0.05). Across four complex scenarios, it exhibits stronger robustness than mainstream models (YOLOv5n-seg, YOLOv8n-seg, YOLOv10n-seg, YOLO11n-seg), and achieves 17/38 FPS on Jetson Nano B01/Orin. Additionally, when combined with DeepSORT, compared with random image sampling, the mean error between predicted and actual residual film area decreases from 232.30 cm² to 142.00 cm², and the root mean square error (RMSE) drops from 251.53 cm² to 130.25 cm². This effectively mitigates pose-induced random errors in static images and significantly improves area estimation accuracy. Full article

Alkanna tinctoria (L.) Tausch is a valuable medicinal plant known for its root-derived hydroxynaphthoquinone enantiomers, alkannin/shikonin (A/S), which exhibit significant pharmaceutical and cosmeceutical potential. However, its limited natural distribution and overharvesting pose conservation challenges, necessitating sustainable cultivation and extraction strategies. The application of Natural Deep Eutectic Solvents (NaDESs) has garnered significant attention as sustainable alternatives to conventional solvents. However, their toxicity in living plant systems remains largely unexplored. This study presents the successful establishment of an ex situ hydroponic cultivation system using the nutrient film technique (NFT) to grow A. tinctoria under greenhouse conditions. The system promoted plant acclimatization, vigorous root development, and initial production of A/S derivatives. In parallel, the toxicity evaluation of a bio-based NaDES, LeG_5_20 (levulinic acid–glucose, 5:1, with 20% water), applied as a circulating medium, was assessed. Physiological stress responses of the plants to NaDES circulation were assessed through non-destructive measurements, including stomatal resistance, photosynthetic and transpiration rates, and sub-stomatal CO₂ concentration. Short-term (24 min) exposure to NaDES showed no significant adverse effects, while longer exposures (4–8 h) induced marked stress symptoms and loss of leaf area. These findings demonstrate the feasibility of integrating green hydroponic systems with eco-friendly extraction solvents and provide a framework for further optimization of plant age, solvent exposure time, and system design to enable sustainable metabolite recovery without plant destruction. Full article

Accurate estimation of flow velocities in torrents and steep rivers is essential for flood risk assessment, sediment transport analysis, and the sustainable management of water resources. While deep learning models are increasingly applied to such tasks, their evaluation often depends on statistical metrics that may yield conflicting interpretations. The objective of this study is to clarify how different evaluation metrics influence the interpretation of hydrological deep learning models. We analyze two models of flow velocity prediction in a torrential creek in Taiwan. Although the models differ in architecture, the critical distinction lies in the datasets used: the first model was trained on May–June data, whereas the second model incorporated May–August data. Four performance metrics were examined—root mean square error (RMSE), Nash–Sutcliffe efficiency (NSE), Willmott’s index of agreement (d), and mean absolute percentage error (MAPE). Quantitatively, the first model attained RMSE = 0.0471 m/s, NSE = 0.519, and MAPE = 7.78%, whereas the second model produced RMSE = 0.0572 m/s, NSE = 0.678, and MAPE = 11.56%. The results reveal a paradox. The first model achieved lower RMSE and MAPE, indicating predictions closer to the observed values, but its NSE fell below the 0.65 threshold often cited by reviewers as grounds for rejection. In contrast, the second model exceeded this NSE threshold and would likely be considered acceptable, despite producing larger errors in absolute terms. This paradox highlights the novelty of the study: model evaluation outcomes can be driven more by data variability and the choice of metric than by model architecture. This underscores the risk of misinterpretation if a single metric is used in isolation. For sustainability-oriented hydrology, robust assessment requires reporting multiple metrics and interpreting them in a balanced manner to support disaster risk reduction, resilient water management, and climate adaptation. Full article

Accurately predicting evaporation duct height (EDH) is a crucial technology for enabling over-the-horizon communication and radar detection at sea. To address the issues of overfitting in neural network training and the low efficiency of manual hyperparameter tuning in conventional evaporation duct height (EDH) prediction, this study proposes the application of Bayesian optimization (BO)-based deep learning techniques to EDH forecasting. Specifically, we developed a novel BO–LSTM hybrid model to enhance the predictive accuracy of EDH. First, based on the CFSv2 reanalysis data from 2011 to 2020, we employed the NPS model to calculate the hourly evaporation duct height (EDH) over the Yongshu Reef region in the South China Sea. Then, the Mann–Kendall (M–K) method and the Augmented Dickey–Fuller (ADF) test were employed to analyze the overall trend and stationarity of the EDH time series in the Yongshu Reef area. The results indicate a significant declining trend in EDH in recent years, and the time series is stationary. This suggests that the data can enhance the convergence speed and prediction stability of neural network models. Finally, the BO–LSTM model was utilized for 24 h short-term forecasting of the EDH time series. The results demonstrate that BO–LSTM can effectively predict EDH values for the next 24 h, with the prediction accuracy gradually decreasing as the forecast horizon extends. Specifically, the 1 h forecast achieves a root mean square error (RMSE) of 0.592 m, a mean absolute error (MAE) of 0.407 m, and a model goodness-of-fit (R²) of 0.961. In contrast, the 24 h forecast shows an RMSE of 2.393 m, MAE of 1.808 m, and R² of only 0.362. A comparative analysis between BO–LSTM and LSTM reveals that BO–LSTM exhibits marginally superior accuracy over LSTM for 1–15 h forecasts, with its performance advantage becoming increasingly pronounced for longer forecast horizons. This confirms that the Bayesian optimization-based hyperparameter tuning method significantly enhances model prediction accuracy. Full article