Flood forecasting in river networks requires models that can capture both spatial dependencies across river basins and the temporal dynamics of hydrological processes. This study proposes a hybrid approach combining Graph Neural Networks (GNNs) with temporal sequence models, such as Long Short-Term Memory (LSTM) networks and Transformer encoders, to forecast water levels in the Vu Gia-Thu Bon (VGTB) River Basin, a highly flood-prone catchment in Central Vietnam. The models are trained using multi-source historical datasets, including water level observations, rainfall records, and reservoir release data, and are evaluated for their ability to predict water levels several hours in advance. Two model variants are developed: (1) a GNN-LSTM model, which integrates GNN spatial learning with LSTM for temporal encoding, and (2) a GNN-Transformer model, which leverages a Transformer encoder to effectively capture long-range temporal dependencies. Results show that the hybrid GNN-temporal models significantly outperform a conventional LSTM model, with the GNN-Transformer model achieving a 20-30% reduction in root-mean-square error (RMSE). The GNN-LSTM model also demonstrates notable improvements, highlighting the importance of incorporating river network connectivity into flood forecasting. The study demonstrates that the hybrid models are particularly effective during major flood events, as they better capture rapid water level rises and the spatial propagation of floods. The findings suggest that integrating graph-based spatial learning with advanced temporal encoders offers a promising direction for improving flood forecasting accuracy, thereby contributing to more effective flood risk management and water resources planning.

Surface root mean square height (SRMSH) is a key parameter characterizing surface roughness; it reflects soil hydrological properties and influences related physical processes. LiDAR offers an effective means for measuring SRMSH over large areas, yet the method involves several uncertainty factors that require further investigation. In this study, we proposed a method based on UAV LiDAR data, utilizing linear sampling in target areas. Three potential sources of uncertainty are examined: the spatial interpolation method, the interpolation interval, and the linear sampling length. By selecting and combining representative values for these factors, SRMSH is computed for multiple sample areas. The effects of each factor are then evaluated by comparing results of different parameter configurations. The main findings are as follows: (1) The linear sampling method is capable of estimating SRMSH but introduces systematic errors that correlate primarily with the sampling length; (2) Sensitivity analysis reveals that the sampling length dominates measurement outcomes (contributing >50% of variation), followed by the interpolation method (<10%), while the interpolation interval has minimal influence (<2%). The quantification of parameter impacts provides valuable methodological references for optimizing measurement protocols and developing robust alternatives.

Reservoir operations play a pivotal role in modifying drought propagation processes, particularly by influencing the transition from meteorological to hydrological drought. This study investigates the drought propagation characteristics in the middle reaches of the Hanjiang River Basin, China, under both natural and observed (reservoir-influenced) conditions. The Standardized Precipitation Evapotranspiration Index and Standardized Streamflow Index were utilized to characterize meteorological and hydrological drought, respectively. The Soil and Water Assessment Tool was employed to reconstruct natural streamflow, providing a baseline for comparison. A nonlinear copula function was applied to model the dependence between meteorological and hydrological drought characteristics, and a Copula-Bayesian network was developed to quantify propagation probabilities. Under the regulation of the Danjiangkou Reservoir, drought propagation characteristics for 1-12-month timescales have shifted markedly: the average propagation time downstream was prolonged from 0.25-0.70 months to 0.94-2.36 months, while the propagation rate declined from 0.83-0.89 to 0.48-0.65, and the sensitivity decreased from 0.83-0.96 to 0.68-0.79. In the natural scenario, the optimal propagation model was based on the Gumbel copula, whereas the observed scenario was best fitted by the Frank copula. The likelihood of hydrological drought increased with the intensity and duration of meteorological drought. However, compared to natural conditions, reservoir regulation significantly delayed the onset and reduced the probability of hydrological drought occurrence. These findings elucidate the nonlinear dynamics of drought propagation and underscore the regulating effect of large-scale reservoirs on downstream hydrological responses.

Understanding the propagation patterns of hydrological droughts is crucial for drought prevention, disaster mitigation, and water resource management. Two common methodological frameworks are employed: standardized indices, represented by the Standardized Streamflow Index (SSI), and threshold-based (non-standardized) indices, represented by the variable drought threshold (VDT) and fixed drought threshold (FDT). However, differences and similarities between these two types of methods in identifying hydrological droughts and characterizing their propagation patterns (e.g., onset, peak intensity, and termination) have not been systematically examined. To address this gap, the source region of Yellow River basin (SRYB), an area with relatively limited human influence, is selected as a case study. The results reveal several key similarities and distinctions: (i) The average duration of hydrological droughts during 1956-2022 is similar between SSI and VDT, but significantly longer when identified by FDT. The average severity derived from FDT is lower than that from VDT. (ii) All three methods effectively capture the spatial propagation behavior of hydrological droughts across the SRYB, which generally exhibits a decreasing intensity from upstream to downstream. (iii) Marked differences exist in the intra‑annual timing of event occurrences: SSI and VDT show an approximately even monthly distribution, whereas FDT indicates a pronounced concentration of drought events in low-flow periods. (iv) The onset, peak intensity, and termination of hydrological droughts identified by FDT align well with wet-dry transitions, a feature not reflected in the results from SSI or VDT. (v) These discrepancies stem from the “relative” nature of SSI and VDT—which evaluate drought events against reference-period thresholds, leading to similar average durations and uniform seasonal distributions—and the “absolute” nature of FDT, which uses inherent low-flow thresholds, resulting in longer durations, reduced severity, seasonal concentration, and better alignment with wet-dry transitions. This study provides critical and actionable insights for selecting appropriate hydrological drought identification methods, thereby supporting sustainable water resource management and enhancing water security under drought conditions.

Microplastics are widely distributed in aquatic environment and serve as habitats for microbial communities, which pause threats to the ecosystem. In this study, the seasonal variations in microplastic abundance and biogeography of their associated plastisphere were studied from 14 sites along the Wuhan Reaches of Yangtze River across spring and autumn. Significant variability in microplastic abundance was observed in Wuhan reaches of Yangtze River, ranging from 260 to 1540 items/m³. The microplastic concentrations were significantly higher in autumn than those in spring. The sizes, colors, and shapes of microplastics detected in two seasons were dominated by less than 1000 µm, black and transparent, and fibers and fragments, respectively. The main composition of microplastic polymers were polyethylene terephthalate (PET), polytetrafluoroethylene (PTFE) and polyethylene (PE). Medium risk pollution load index (PLI), high to extremely high polymer risk index (PRI) and medium risk potential ecological risk index (PERI) were observed, with higher risks in spring than in autumn. Significant seasonal differences in richness and composition of bacterial communities in water and plastisphere observed, with higher richness in spring. Proteobacteria and Actinobacteriota were the dominant phyla in both water and plastisphere. Stochastic processes dominated the assembly processes of bacterial communities. Plastisphere had less complex co-occurrence networks than water, with denser networks in spring than autumn. Environmental factors influenced the composition of bacterial communities, with total nitrogen (TN), total phosphorus (TP), total organic carbon (TOC) and longitude having strong significant effects. This study shed light on the microplastic pollution and their associated bacteria in the Wuhan reaches of Yangtze River.

An essential part of river ecosystems, dissolved organic matter (DOM) has an enormous effect on the transport and variability of heavy metals. Nonetheless, our understanding of how the DOM affects heavy metal behavior in agricultural rivers remains limited. In this research, the DOM characteristics and effects on heavy metals (Cu²⁺, Zn²⁺, and As³⁺) were analyzed in the Liaocheng section of the Tuhui River (LCTH) during High, Normal, and Low-water periods. EEM-PARAFAC identified three fluorescent components: C1 are microbial humic-like substances, C2 are terrestrial humic-like substances, and C3 are protein-like substances. Humic-like substances were dominant in all water periods, accounting for approximately 70% of the total DOM. It is interesting to note that the ratio of microbial-derived components to terrestrial-derived components, (C1 + C3)/C2 in Chiping District (3.30 ± 1.18) was significantly higher than that in Shen County (2.40 ± 0.57). On this basis, to further substantiate the binding capacity of various types of DOM with Cu²⁺, samples from Lao Nanzhen (LNZ) and Li Fengtao (LFT) sites were selected for fluorescence titration experiments. The results showed that the samples from both sites were humic-like substances preferentially complexed with Cu²⁺. The complexation of terrestrial humic substances (C2) with Cu²⁺ was significantly higher in LFT (logK_a=3.65±0.08) than in LNZ (logK_a=3.16±0.06). Structural equation modeling (SEM) revealed that humic-like substances in the LCTH had a greater effect on different heavy metals than protein-like substances. The findings elucidate how DOM affects heavy metals, contributing to efforts aimed at controlling heavy metal pollution in river systems.

This study was conducted to address the concerns over the changes in river morphology and their implications for sustainable river management. It investigates the morphological dynamics of the Upper Jhelum River between 2018 and 2023, using high-resolution Sentinel-2 imagery and advanced geospatial techniques to capture subtle yet significant changes in river morphology during the period. Remote sensing techniques integrated with MATLAB-based analytical algorithms were used to delineate and quantify changes in river planform, width, migration, and spatial patterns of erosion and deposition. This provided comprehensive evidence of the river's adaptive responses to both natural and anthropogenic influences. Results indicate substantial variations in the morphodynamic parameters, with river width ranging from 590.15 m (at 125 km) to 34.8 m (at 117.8 km), an annual centerline migration rate of 5.36 m, and distinct erosion-accretion dynamics—20.38 km² of accretion upstream (Segment A) and 26.94 km² of erosion downstream (Segment C). The Flow Duration Curves (FDCs) for 2018 and 2023 focus on the role of discharge variability in driving these morphological changes, with higher flows in 2018 corresponding with increased width and channel migration compared to 2023. Based on these findings, targeted management interventions were proposed for this river system. Thus, this study, conducted by integrating remote sensing technology with automated analytical algorithms, enhances our understanding of fluvial dynamics and is crucial for river restoration and management. This study establishes a critical baseline for future research on the Jhelum River's morphodynamics and human-induced modifications.

To accelerate the development of impulse hydro-turbines and support the efficient utilization of hydropower resources in Southwest China, this paper examines the background, historical development, and research progress of impulse hydro-turbines. By analyzing energy development data, hydropower potential, and research trends in impulse hydro-turbines, we provide a comprehensive review based on existing studies and technical achievements from domestic and international scholars. This paper focuses on five key components: distributors, injectors (including deflectors), runners, auxiliary systems, and engineering applications. For each part, the working principles and technological progress are detailed. Furthermore, five main findings are highlighted—such as the role of impulse hydropower in addressing China's energy crisis—and five recommendations are put forward, including the need to strengthen related technological capabilities in China. This review aims to provide a reference for further research and industrial development.

This study presents an integrated flood forecasting and inundation mapping framework using coupled SWAT, HEC-HMS, and HEC-RAS models for the Muga River basin in the Abay Basin, Ethiopia. The study addresses the limited application of integrated hydrologic-hydraulic modeling and the uncertainty associated with digital elevation model (DEM) resolution in data-scarce regions. Flood discharges corresponding to 2-, 5-, 10-, 25-, 50-, and 100-year return periods, which are commonly used for flood risk assessment and infrastructure design, were simulated. SWAT was applied for continuous hydrological simulation, while the Hydrologic Engineering Center's Hydrologic Modeling System (HEC-HMS) supported event-based peak flow estimation. Model performance was evaluated using observed streamflow and the 2017 historical flood event. Flood inundation characteristics were assessed using ALOS 30 m, SRTM 30 m, and SRTM 90 m DEMs to quantify terrain-related uncertainty. Results show that SWAT produced slightly higher peak discharges than HEC-HMS, providing conservative estimates for flood risk planning; therefore, SWAT-derived flows were used for final inundation mapping. Among the tested DEMs, ALOS 30 m exhibited the closest agreement with observed flood extent and depth. The proposed framework is transferable to other sub-basins of the Abay Basin with similar hydro-geomorphic characteristics and offers practical support for flood hazard mitigation and planning.