Effective management of multi-purpose reservoirs requires precise planning and accurate data to balance competing objectives and constraints. Reservoir inflow forecasting is critical in this process, with deep learning models increasingly applied across various time scales, from hourly to annual predictions. This study integrated of a two-layer stacked long short-term memory network with decomposed data and a rolling window technique to enhance multi-day reservoir inflow forecasting accuracy. The proposed framework was applied to the Lam Takhong Dam in northeastern Thailand, a tropical monsoon region characterized by distinct wet and dry seasons. The dataset included daily reservoir inflow, river discharge, and average rainfall records spanning multiple years. Four forecasting strategies were compared for up to 7-d predictions: multi-step prediction, rolling prediction, multi-step prediction with decomposition, and rolling prediction with decomposition. The results indicated that while all models performed similarly for short-term predictions, accuracy declined over longer forecasting horizons. The rolling window approach with decomposition consistently outperformed others, achieving an average correlation coefficient of 0.92 and an average Nash–Sutcliffe model efficiency coefficient of 0.78 at the 7-d forecasting horizon. These findings demonstrate the practical advantages of integrating decomposition into a dynamic forecasting framework, particularly in reducing error accumulation in extended hydrological predictions.

Textile wastewater contains recalcitrant dyes and organics that are difficult to degrade via conventional treatments. This study evaluated the reduced graphene oxide (rGO)-supported nanoscale zero-valent iron (nZVI) composite (rGO/nZVI) for treating real textile wastewater in batch and continuous systems. The rGO/nZVI catalyst was synthesized and characterized using scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), and Brunauer–Emmett–Teller (BET) analyses, confirming uniform iron dispersion, active functional groups, and a mesoporous structure. Batch experiments under varying pH (3.0–5.5), catalyst dosages (150–1 000 mg/L), and H 2 O 2 concentrations (150–1 000 mg/L) identified optimal conditions (pH of 3, 750 mg/L of rGO/nZVI, 1 000 mg/L of H 2 O 2 , and a reaction time of 110 min), achieving 81.5% chemical oxygen demand (COD) removal (from 450.8 mg/L to 83.5 mg/L) and approximately 90.0% color reduction (from 355–473 platinum–cobalt units (PCU) to 31.9–38.5 PCU). The packed-bed column tests achieved 77.4% COD removal (from 452.4 mg/L to 102.3 mg/L) and approximately 88.0% color reduction (from 362–488 PCU to 42.1–51.8 PCU), demonstrating stable continuous performance. Reusability tests demonstrated catalytic durability over five cycles, with COD removal decreasing from 94.6% to 51.4% and color removal from 96.2% to 65.1%. Overall, rGO enhanced nZVI dispersion, stability, and catalytic activity, supporting rGO/nZVI as a scalable advanced oxidation technology for textile wastewater treatment.

Water sources in volcanic regions have long been a focal point in hydrogeology. Tianchi Lake of the Changbai Mountain in Northeast China, the world's highest volcanic lake, has historically faced water imbalance issues. This study offered a comprehensive analysis of the water sources of Tianchi Lake, examining water volume, hydrodynamics, hydrochemistry, and isotopic evidence. Flow simulations of the Changbai Mountain waterfall during the glacial period indicated that besides local precipitation stored within the mountain during the non-freezing period, other groundwater sources were involved. Additionally, the volume of spring water and the geological structures in the Tianchi Lake area suggested that even expanding the watershed boundary cannot fully account for water balance within the region. Comparative analysis of hydrogen and oxygen isotopes in groundwater and local precipitation within the Changbai Mountain region revealed that external water recharged Tianchi Lake via deep circulation, sustaining the stable flow of Tianchi Lake and its surrounding springs. This study provides valuable insights into the mechanisms and recharge processes of groundwater circulation in volcanic regions.

Turbulent flow around bluff bodies like square cylinders involves complex vortex shedding and flow separation, challenging traditional computational methods. This study developed a novel approach using a generative artificial intelligence (GenAI) model to predict turbulent flow over a single square cylinder. The GenAI model was trained using high-fidelity simulation data generated from an advanced differentiable physics framework (PhiFlow), which can efficiently capture the nonlinear dynamics of turbulent flow. Flow predictions from the GenAI model were validated against numerical results, demonstrating high accuracy in capturing key flow characteristics, including vortex shedding frequency. Stability and spatial–temporal frequency analyses revealed strong agreement between the diffusion model and numerical simulations. This study highlights the potential of GenAI models to significantly enhance the prediction and analysis of turbulent flow, offering a powerful tool for fluid dynamics research and engineering applications.

The Three Gorges Reservoir (TGR) is one of the largest hydroelectric projects in the world, with significant impacts on the hydrology and ecology of the Yangtze River Basin. Understanding the effects of TGR operation on surrounding water systems, especially the Jingjiang Reach and Dongting Lake, is crucial for local water resources management and flood control. This study evaluated the impact of the TGR on water diversion in the Jingjiang Reach and outflow from Dongting Lake using observed data and sedimentation patterns before and after TGR operation. A coupled one-/two-demensional hydrodynamic model was developed to simulate hydrological processes. The relationship between TGR scheduling and Dongting Lake inflow and outflow across different periods was quantified. The results indicated that after TGR operation began, riverbed erosion significantly lowered tributary water levels under equivalent main stream flow. Lake inflow through the three Jingjiang Reach outlets increased during drawdown and water supplement periods but decreased during flood and impounding periods. Lake outflow increased during drawdown, flood, and water supplement periods but declined significantly during the impounding period. The contributions of factors varied considerably. Reservoir scheduling accounted for 328.86% of inflow changes at the Taiping outlet during the drawdown period but only 20.72% during the flood season. River topography changes contributed 157.41% to lake outflow changes during the drawdown period, but only 1.85% during the water supplement period. These findings enhance our understanding of river–lake system evolution and support improved management strategies.

With over 1.3 billion people worldwide facing irregular water access, efficient water management is a global priority. This study presented a comprehensive approach for optimizing the operation of intermittent water distribution networks through the creation of district metered areas (DMAs). It advanced traditional DMA design by integrating network partitioning with optimized operational schedules, offering a practical framework for managing intermittent water supply systems. The proposed methodology aims to reduce water losses while improving service equity and quality. First, the network is partitioned using the fast-greedy community detection algorithm based on modularity from graph theory, enabling DMAs to operate independently at different times of a day. Flow control valves are installed at DMA entry points, while isolation valves isolate remaining boundary pipes, enhancing operational flexibility. Second, the particle swarm optimization algorithm optimizes the operational schedule of each DMA and determines the optimal start time and water supply duration for each DMA. This step minimizes total daily distributed volume while ensuring adequate service. This approach reduced the daily distributed volume of the Modena network by approximately 720.0 m 3 and significantly decreased the leakage rate from 30.5% to 18.7%, demonstrating its effectiveness.