Journals
2026 EN
Li Yingrui · Zuo Qiting · Dou Shentang
+1 more
In modern rainfall monitoring, forecasting, and early warning systems, radar serves as a critical technological foundation, profoundly transforming approaches to flood management and disaster mitigation. To improve the accuracy of radar-based rainfall retrievals and effectively assess the reliability of the results, this study introduces an innovative, transferable three-stage radar–rainfall retrieval framework (TS-RR), consisting of three essential components: matching, retrieval, and optimization. Additionally, a novel point-real-process (PAP) three-dimensional evaluation method is proposed, enabling a comprehensive and quantitative assessment of retrieval accuracy from three perspectives: stations, spatial distribution, and rainfall processes. The results demonstrate that (1) the TS-RR framework effectively guides radar–rainfall retrieval, with the optimization phase (Stage III) significantly enhancing the capture of rainfall peaks, improving spatial distribution accuracy (KLD < 0.1), and demonstrating strong adaptability to dynamic rainfall processes (CC > 0.9). (2) Under limited sample conditions, the PMM in Stage II exhibits strong robustness, maintaining stable correlation performance (correlation coefficient, CC = 0.7–0.9). (3) The PAP approach effectively evaluates the reliability of the retrieval results. The proposed method offers practical value by providing scientifically grounded data support and decision-making references for water resource management, flood forecasting, and early warning systems.
Journals
2026 EN
Zhang Cong · Wan Hongmei · Tang Jingru
+2 more
The Himalayan region is distinguished by a high density of glacial lakes and frequent seismic activity. Under seismic actions, glacial lakes are prone to produce large-amplitude surge waves, potentially leading to the overtopping failure of the glacial lake dams. This study systematically investigates the seismic-induced water waves (SIWWs) of glacial lakes through shaking table experiments and OpenFOAM-based simulations. The key findings are as follows: (1) The response of SIWWs is primarily controlled by seismic frequency, peak ground acceleration, and water depth, with little sensitivity to seismic duration. (2) Building upon these controlling parameters, a calculation formula is developed to estimate the maximum SIWW height, which is further combined with hydraulic thresholds for sediment mobilization to establish a critical criterion for glacial lake overtopping failure. (3) By integrating regional probabilistic seismic hazard levels with remote sensing data, a risk classification framework for glacial lake failure under SIWWs is proposed and demonstrated through a case study of the Gyirong Valley in the central Himalaya. This remote sensing-based framework provides a practical tool for assessing glacial lake failure risk under earthquakes and contributes to GLOF risk mitigation in the Himalayan region.
Journals
2026 EN
Chen Xincong · Ji Chunning · Xu Dong
+2 more
Conventional studies of vegetated shear layers have typically modelled aquatic vegetation as rigid, upright structures to simplify submerged canopy hydrodynamics. While this approach has advanced the understanding of basic flow-canopy interactions, it overlooks vegetation bending and swaying, which are critical but challenging to characterize in flexible systems. As a result, comparative analyses of turbulent flow dynamics between flexible and rigid canopies remain limited. This study employs a coupled fluid-structure interaction framework to investigate turbulent open-channel flows across four vegetation configurations: (1) rigid upright canopy, (2) rigid curved canopy, (3) flexible canopy with constrained motion, and (4) flexible canopy with unconstrained sway. The hydrodynamics are resolved using Large Eddy Simulation, while vegetation motion is computed with the Vector-Form Intrinsic Finite Element method. This integrated approach enables systematic evaluation of how vegetation compliance modifies momentum transport. Results show that dynamic sway generates additional flow resistance beyond bending reconfiguration, amplifying shear-layer turbulence relative to rigid canopies. Under strong reconfiguration, unconstrained flexible canopies exhibit mean sway amplitudes approaching half the stem diameter, producing up to 60% enhancement in Reynolds stresses and 20% increase in turbulent kinetic energy within the canopy. Streamwise bending and sway also enhance mechanical dispersion, with dispersive stresses 10–50% greater than Reynolds stresses, and near-bed stress ratios peaking at 400%. Flow visualization reveals that streamlined bending suppresses flow-around effects and promotes symmetric secondary circulation, whereas lateral sway expands turbulence propagation and intensifies mixing. Momentum exchange is dominated by coherent Kelvin-Helmholtz vortices at the canopy-water interface, which transfer momentum 5–6 times more efficiently than other turbulent mechanisms. Streamlined reconfiguration reinforces turbulence organization, while multi-directional sway shifts transport deeper into the canopy, enhancing diffusion. These findings highlight the critical role of vegetation flexibility in shaping canopy hydrodynamics and provide new insights into turbulence-vegetation interactions in aquatic environments.
Journals
2026 EN
Wang Zhouhan · Hong Zhifeng · Chen Junfeng
+12 more
Since July 2025, Foshan has experienced the largest outbreak of Chikungunya fever in China. This single-centre retrospective study included 1,908 laboratory-confirmed cases. Phylogenetic analysis identified the East/Central/South African lineage, closely related to the 2025 La Réunion strain. The included cases were all non-severe, showed no sex bias, but exhibited clear regional clustering. Fever (86.8%), arthralgia (85.4%), and rash (64.1%) were predominant symptoms. Fever and rash were more common in minors, while arthralgia increased with age. Lymphopenia occurred in 65.5% of patients and was more common in elderly patients, who had higher viral loads and longer RNA clearance times. Age, sex and viral load independently influenced clinical manifestations and laboratory characteristics. The findings provide initial descriptive evidence and highlight age-related differences in chikungunya's clinical and virological profiles. Long-term follow-up and larger-scale investigations were necessary to provide evidence for clinical decision-making.
Journals
2026 EN
Dou Hongwen · Zhang Kun · Zmeureanu Radu
This paper presents the development and validation of grey-box models for estimating the chilled water temperature difference ΔT chw across the chiller evaporator, with potential applications as virtual sensors in building automation systems (BAS) or integration into other mathematical models. The models are established for two scenarios, variable and constant chilled water flow rates under quasi-steady-state operation. These models require a small number of input variables and are characterized by strong adaptability. Three case studies of different chillers are used to validate the proposed virtual sensors with both static and dynamic windows methods, and help in the generalization of the proposed method. The models demonstrate high accuracy and robustness, achieving a root-mean-squared error of 0.19 °C in one case study. This study addresses the gap in the availability of simple yet reliable models that can be practically integrated into building automation systems for virtual sensing, virtual calibration, fault detection and diagnosis, and HVAC system control and optimization.
Resource
2026 EN
Fengyuan Dong · Xing Xu · Zeyang Dou
Power drone inspections have become a crucial method for monitoring the status of transmission lines and substation equipment. However, in real-world scenarios, defect targets typically exhibit features such as small scale, weak texture, and complex backgrounds with strong interference. The traditional model struggles to meet the demand for rapid on-site closed-loop resolution. To address these challenges, this paper proposes an on-site defect identification method for drone inspection scenarios based on spatial reasoning learning from multimodal image-text large models. First, we develop a defect image generation method using Variational Autoencoders (VAE) and conditional score matching diffusion. By embedding semantic information into the feature layer of the U-Net decoder via the multi-layer Spatial Adaptive Normalization and Decomposition (SPADE) operator, we precisely control the generation location and morphology of defect features. Second, at the model level, a visual encoder based on Swin Transformer is constructed. A hierarchical window attention mechanism extracts multi-scale defect and scene topological features. A text encoder is built using the Next Token Prediction pre-training method and a large language model, combined with domain-specific fine-tuning using power system terminology databases and equipment ledger knowledge. At the cross-modal fusion layer, a contrastive learning mechanism aligns defect images and textual descriptions within a unified vector space. For spatial reasoning learning, we construct visually-linguistically intertwined regional inference samples. Region-level instruction fine-tuning drives the model to perform dynamic region cropping and multi-step reasoning analysis. Regarding on-site deployment, we combine CPU-NPU heterogeneous co-acceleration with memory optimization strategies. Through dynamic task allocation and data prefetching techniques, we achieve low-power, low-latency real-time edge inference. Experimental results demonstrate that our method delivers high-precision and high-efficiency defect detection in power inspection scenarios, effectively addressing complex backgrounds and resource-constrained field environments.
Resource
2026 EN
Yexin Dou · Haijiang Wang · Jian Wan
+3 more
One-shot federated learning (FL) completes model training and aggregation in a single communication round, significantly reducing communication costs compared to traditional FL. This approach is particularly suitable for resource-constrained environments such as wireless sensor networks (WSNs). However, existing solutions face significant challenges in aggregation owing to model heterogeneity, where clients adopt architectures of varying depth, width, and computational capacity. To address this issue, we propose a one-shot FL method named FedLIM, which employs a lightweight intermediate model for efficient knowledge transfer and global model aggregation. The Fisher information matrix (FIM) is incorporated to guide the model aggregation process and improve its robustness. Although FedLIM completes global training and aggregation in a single communication round, an optional personalized model adjustment step is introduced afterward. This step only involves server-to-client distribution without additional aggregation. Experimental results on three datasets demonstrate that FedLIM achieves superior global model accuracy compared to existing one-shot FL methods, particularly in highly heterogeneous environments. Moreover, the accuracy of local models is further enhanced through this optional refinement step.
Resource
2026 EN
Huadong Guo · Changyong Dou · Dong Liang
+13 more
The implementation of the United Nations (UN) 2030 Agenda for Sustainable Development (2030 Agenda), with its 17 sustainable development goals (SDGs), faces challenges such as insufficient data, limited research methodologies, and uneven progress across regions. Earth observation (EO), particularly scientific satellites, offers unique advantages in supporting global sustainable development by providing objective, dynamic, and large-scale datasets for SDG evaluations and policymaking, as well as by facilitating the study of Earth’s environmental systems and their interactions with human activity. Sustainable Development Science Satellite 1 (SDGSAT-1), the world’s first scientific satellite dedicated to supporting the 2030 Agenda, was designed and developed by the International Research Center of Big Data for Sustainable Development Goals (CBAS). Its three advanced EO sensors, i.e., Thermal Infrared Spectrometer (TIS), Glimmer Imager (GLI), and Multispectral Imager (MSI), furnish high-quality data, enabling continuous monitoring of human activity and environmental changes to bolster SDG-related research and global sustainability initiatives. As of November 2025, SDGSAT-1 has collected over 480 000 global terrain coverage images since its launch in November 2021. All its datasets have been shared free of charge with the Global Scientific Community through the SDGSAT-1 Open Science Program initiated in September 2022. The datasets have enabled researchers from more than 110 countries, 10 UN agencies, and various international organizations to publish over 180 scientific articles, 17 UN reports, and numerous public data products. These have demonstrated applications in urban development, disaster response, environmental monitoring, agriculture, and marine conservation. This article reviews the technical innovations and mission specifications of the SDGSAT-1 satellite, demonstrates its contribution in leveraging space technology for SDG monitoring and evaluation, and discusses the future evolution of development of EO systems, specifically the planned Sustainable Development Satellite Constellation, for supporting the global achievement of SDGs.
Resource
2026 EN
Qianlong Yang · Xiaozhen Qiao · Yonggang Du
+3 more
Hyperspectral image classification demands models capable of efficiently capturing complex spectral–spatial relationships and long-range dependencies. Despite significant advances in CNNs and Transformers, balancing modeling capacity with computational efficiency remains challenging. Recently, Mamba has emerged as a promising alternative with linear-complexity global modeling capabilities. However, current Mamba-based methods suffer from two critical limitations: artifact tokens that compromise foreground-background distinction, and overfitting tendencies exacerbated by their non-hierarchical, homogeneous layer architecture. To address these limitations, we propose PGSMamba, a prompt-guided shuffle state space model for hyperspectral image classification. Our approach incorporates three core components: a patch-prompt sequence construction module for artifact filtering through learnable prompts, a shuffle scanning module providing layer- wise regularization via random perturbations, and a prompt-to-patch aggregation strategy enabling robust global-local feature fusion and effectively suppressing artifact tokens within patch tokens. Comprehensive experiments on five benchmark datasets, including Indian Pines, Pavia University, Houston 2013, WHU-Hi-LongKou, and Salinas, demonstrate that PGSMamba achieves state-of-the-art overall accuracies of 97.88%, 98.09%, 94.02%, 98.92%, and 95.11%, respectively. These results confirm the effectiveness and generalization ability of our approach, highlighting PGSMamba as a promising solution for efficient and robust hyperspectral image classification.
Resource
2026 EN
Yi-Huan Ou-Yang · Hsiao-Chun Lin · Chi-Wei Huang
+3 more
Goal: Existing beta burst detection algorithms for closed-loop deep brain stimulation (DBS) are computationally complex, limiting their use in implantable devices. We aimed to develop an improved beta burst extraction algorithm for chip-based DBS devices with real-time model updating. Methods: Building on an established beta burst detection method, we proposed a sliced mechanism for peak frequency finding and modified burst extraction for information sharing with real-time model updating. Results: Testing on rat electrocorticographic (ECoG) recordings showed that the proposed algorithm maintains a strong correlation (ρ = 0.89 ± 0.06) with the conventional method, with a 53.3% reduction in computational complexity for peak frequency finding. Conclusions: Integrating this improved beta burst detection into chip-based DBS devices represents a key algorithmic advancement toward adaptive neuromodulation therapies. The strong correlation and reduced complexity validate our proposal for real-time neural biomarker tracking, facilitating hardware and chip implementation, and advancing the development of implantable systems.