Accurate measurement of river surface flow velocity is essential for water disaster prevention, water resources management, and hydrological research. While optical flow methods based on image velocimetry have garnered significant interest for their non-contact advantages, they remain prone to substantial estimation errors when applied to natural river environments characterized by complex surface motion patterns, pronounced non-rigid deformations, and multiscale nonlinear flow dynamics. To address these challenges, this study proposes a multi-scale local–global deep optical flow network (MSKFlow) for river surface velocity measurement via drone video imagery, aiming to improve measurement accuracy in complex water surface environments. The framework first introduces a joint feature extraction strategy that integrates multi-scale convolution with the Swin Transformer, which effectively enhances feature representation and matching capabilities under complex flow field conditions. Moreover, a superkernel updater is employed to replace the conventional convolutional gated recurrent unit, significantly improving optical flow refinement in complex flow field structures. Experimental results based on a synthetic river dataset demonstrate that MSKFlow can accurately reconstruct flow field structures under controlled conditions. Field validation using drone-recorded videos from two reaches of the Heihe River Basin shows that, when benchmarked against current meter measurements, the estimated surface flow velocities yield average relative errors of 14.4% and 15.2%, respectively. These results confirm the reliability and stability of the proposed method under natural river surface conditions. Overall, MSKFlow offers an efficient and promising solution for drone-based remote sensing monitoring of river surface flow velocity in complex environmental settings.

Employing K-means clustering to reconstruct missing water surface elevations in LiDAR digital elevation models for hydrodynamic simulation.

Tornadoes can cause extensive destruction to the built environment, yet limited research in China has examined how building attributes and surrounding natural environments influence the vulnerability of rural buildings. Using the “6.23” tornado in Yancheng City, Jiangsu Province (China), this study applied Logistic Regression and eXtreme Gradient Boosting (XGBoost) models to analyze these relationships. A dataset of 2521 rural buildings shows that four building attributes, such as number of stories, structural type, usage, and floor area, were significantly associated with tornado damage to rural buildings at the 95% confidence level. The results suggest that taller buildings, larger areas, and stronger structures like brick-concrete and masonry are generally less vulnerable. Beyond building-level factors, geographic context and surrounding land use also played a critical role. To capture these broader effects, individual buildings were aggregated into 42 villages, and average village-level damage was calculated. Modeling at the village scale reveals that higher proportions of arable land (e.g., dry farmland), built-up land (e.g., rural settlements), and river surfaces tended to increase vulnerability, whereas transportation land had a mitigating influence. The consistency of results across Logistic Regression and XGBoost strengthens the reliability of these findings. Overall, this study highlights the combined importance of building design and environmental context in shaping rural building vulnerability to tornadoes. By providing data-driven evidence, the results support local governments in developing adaptation planning strategies for tornado-prone rural areas, particularly through informed decisions on building attributes and land use configurations to reduce future risks.

Quantifying river surface flow velocity is essential for hydrodynamic modelling, flood forecasting, and water resource management. Traditional in situ methods provide accurate point measurements but are costly and limited in spatial coverage. Unmanned aerial vehicles (UAVs) offer a flexible, non-contact alternative for high-resolution monitoring. Optical flow is a tracer-independent technique for deriving velocity fields from RGB video, making it well suited to UAV-based surveys. However, its operational use is hindered by the limited availability of annotated datasets and by instability under low-texture or noisy conditions. This study combines a Finite element method (FEM)-based physical flow model with UAV video to generate reference datasets and introduces a modified Recurrent All-Pairs Field Transforms (RAFT) architecture based on multiframe sequences. A Gated Recurrent Unit fusion module (Fuse-GRU) is incorporated prior to correlation computation, improving robustness to illumination changes and surface homogeneity while maintaining computational efficiency. The proposed model delivers stable, physically consistent velocity estimates across multiple rivers and flow conditions. Accuracy improves with higher spatial resolution and moderate temporal spacing. Compared to field measurements, the average angular difference ranged from 8 to 15°. The high error values were mainly caused by inaccuracies in the physical model and by complex river features. These findings confirm that multiframe optical flow can reproduce realistic river flow patterns with accuracy comparable to physically-based simulations, thereby supporting UAV-based hydrometric monitoring and model validation.

RivP-RAFT: A patch-based RAFT model for efficient river surface velocity estimation using images

RemoteWaterNet: a lightweight and efficient algorithm for remote river surface velocimetry

Imbricated gravel beds have a distinctive surface morphology, which strongly affects riverbed stability and near-bed hydrodynamic processes. High-resolution digital elevation models combined with image processing techniques were used to quantitatively characterize imbricated surface features across three representative subregions of a gravel bar in the upper Yangtze River. The results indicate that (1) the long axes of imbricated pebbles are predominantly perpendicular to the flow direction; (2) greater bed-surface coarsening corresponds to steeper slope angles, more pronounced imbrication structures, and higher surface roughness; and (3) the self-similarity of the bed surface is anisotropic, being the strongest along the flow direction, and the Hurst index exhibits a quadratic relationship with the inclination index. A field-based method for identifying imbricated clusters was proposed. The results show that where imbrication is more pronounced, clusters tend to form more linear arrangements and increasingly align with the flow direction. These findings provide a new perspective for understanding the microtopographic features of gravel bed river surfaces.

Recent studies have identified the environment as a key reservoir from which antibiotic resistance genes (ARGs) can be acquired and transmitted to pathogens. However, our knowledge about the presence of ARGs in high-flow river sediments is still limited. We analyzed the resistome of sediment bacterial communities along the Swedish river Göta Älv and investigated the potential dissemination of ARGs and antimicrobials from effluents of wastewater treatment plants (WWTPs). While we detected nine different antimicrobials in the effluent water from the WWTPs through HPLC-MS, their presence was not observed in the river surface water. Analysis by qPCR revealed that the genes sul1 and ermB were the most dominant ARGs among sediment, sludge, and effluent samples. Shotgun metagenomics revealed unique differences between the sludge resistomes of the WWTPs. Moreover, our findings show that ARGs increase downstream of the Göta Älv and their diversity differs from that of the upstream sites. Efflux pump resistance-related genes were most abundant in sediment samples, and beta-lactams and tetracyclines were the most common antibiotic classes targeted by ARGs. Our study emphasizes the importance of urban river sediments as a reservoir of ARGs, as tracking ARGs in WWTPs and their receiving environments improves our understanding of their spread and characteristics.

Reliable, real-time river flow monitoring is essential for disaster prevention, but traditional in situ methods are costly and high-risk. Large-scale particle image velocimetry (LSPIV) offers a non-contact alternative, though its accuracy is often compromised by noise and non-water pixels, requiring intensive manual data processing. This study proposes an integrated framework for enhancing non-contact river surface velocity estimation by combining deep learning-based water surface segmentation with optimized LSPIV, using accessible smartphone imaging. The framework was tested on two urban rivers in Taichung, Taiwan. DeepLabV3+ was identified as the superior segmentation model based on MPA/PA and MIoU metrics. The DeepLabV3+-derived mask was integrated into the LSPIV workflow, which was optimized using a 32 × 32 pixels interrogation area (IA), reducing processing time by approximately 44%. By removing non-water pixels, the masked LSPIV yielded a 7% increase in mean surface velocity. This suggests that the inclusion of non-water elements diluted the average, underscoring their tendency to introduce a low-velocity bias in unmasked calculations. The overall validation showed mean absolute percentage errors below 6% compared to the radar velocimeter. Consequently, this integrated smartphone-based framework offers a cost-effective and precise solution for future large-scale deployment in urban flood monitoring and smart city hydrological management.

Accurate estimation of river surface velocity is essential for hydrological monitoring and flood management. However, conventional Large-Scale Particle Image Velocimetry (LSPIV) is often affected by errors arising from inaccurate Region of Interest (ROI) delineation and interference from floating objects or vessels. To overcome these limitations, this study integrates LSPIV with two deep learning models, SegNet and YOLOv8, to enable automated ROI segmentation and vessel detection. SegNet performs real-time identification of water body regions, while YOLOv8 detects and removes vessel intrusions within the ROI, thereby enhancing the precision of velocity estimation. Six field experiments were conducted to assess the performance of the proposed system. The deep learning-enhanced LSPIV achieved Root Mean Square Error (RMSE) values ranging from 0.048 to 0.11 m/s and Normalized RMSE (NRMSE) values between 3.53% and 10.34%, with coefficients of determination (R2) exceeding 0.895 when compared with Acoustic Doppler Current Profiler (ADCP) measurements. SegNet-based ROI segmentation reduced RMSE by up to 0.046 m/s andNRMSE by up to 3.44%, and improved R2 by up to 0.012, while image enhancement further improved segmentation accuracy under varying illumination conditions. Moreover, YOLOv8 successfully detected all vessel intrusions observed in this study, thereby reducing the discrepancies between LSPIV and ADCP-derived velocities from 0.032–0.345 m/s to 0.022–0.314 m/s. Overall, the integration of LSPIV with SegNet and YOLOv8 establishes a highly automated and accurate framework for river surface velocity estimation, demonstrating strong potential for real-time hydrological monitoring and flood risk assessment.

Abstract. The grain sizes of sediments in channels have been linked to landscape characteristics, such as flow distance from headwaters, topographic relief, lithology and climate, in landscapes with little past or present glacial influence. Few studies have explored the controls on sediment characteristics in formerly glaciated landscapes. In this study, we document river surface grain sizes at 279 localities across Scotland. We collect photographs of gravel bars through a citizen science survey, Scotland's Big Sediment Survey. Grain sizes distributions are extracted from the photographs using both manual and automated techniques. We investigate whether grain sizes can be correlated and predicted from environmental variables (e.g., basin slope, flow distance from headwaters) through Spearman's correlation statistics and random forest regression modelling. In contrast to other studies that have primarily focused on non-glaciated landscapes, we find no apparent controls on surface grain sizes in channels across Scotland. Specifically, we find no significant Spearman's relationships between d84 and environmental variables; the strongest relationship was found between d84 and average basin aridity with a weak r2 value of 0.34. We also find that the predictability of our random forest model is poor and only captures 20 % of the variance of d84. We find no correlation between grain size and flow competence, which suggests that sediment is both transport-limited and supply-limited. We propose that Scotland's post-glacial legacy drives the lack of sedimentological trends documented in this study, and that changes in landscape morphology and sediment sources caused by glacial processes lead to a complete decoupling between fluvial sediment grain size and environmental variables. This interpretation aligns with other studies that have highlighted the ongoing role of the post-glacial legacy on landscape evolution in tectonically quiescent terrains, both in Scotland and globally. Our results suggest that fluvial sediment grain size cannot be predicted by a global model based on environmental variables in post-glacial landscapes.

Improving river surface flow velocity measurement by coupling optimal search line algorithm with space-time image velocimetry

Abstract This study proposes an improved FFT-based Space-Time Image Velocimetry (STIV) method for accurate river surface velocity estimation under complex environmental conditions. Based on a comprehensive dataset of real STIs and their corresponding Fourier spectrum images from diverse river scenes, a YOLO-FSD object detection model is developed to identify valid flow signals in the frequency domain. The radius of the bounding box's minimum enclosing circle is used as the angular search radius in FFT-STIV for estimating the Main Orientation of Texture (MOT). Invalid velocity lines are interpolated using cross-sectional velocity distribution. The model achieves 97.70% Precision and 96.91% Recall on the validation set, representing a notable improvement over the baseline model. In two representative cases, the proposed method reduces the average relative error of FFT-STIV from 49.47% to 7.48% and from 91.88% to 6.15% respectively. By combining the powerful feature extraction capabilities of deep learning with the high resolution and interpretability of FFT, this method demonstrates superior robustness and measurement accuracy.

Sufficient textural features are essential for river surface velocity measurements using image velocimetry. However, natural rivers often present a major challenge for accurate image velocimetry due to the scarcity of natural tracers. Developing methods to quickly and efficiently enhance flow features on river surfaces to boost flow velocity measurement precision is of vital important. In this study, we model river surface feature enhancement as a Markov decision process (MDP), and propose a specialized deep reinforcement learning (DRL)-based algorithm for particle dynamic compensation. In order to recognize and eliminate the position deviations in particle compensation resulting from wind interference and the system’s own errors while simultaneously enhancing flow features, we implement dual-frame differencing techniques before and after agent actions as state inputs, and optimize the reward function. We present a novel control framework for particle compensation unmanned aerial vehicles (PCUAV) that uses particle dynamic compensation system (PDCS) to improve river surface feature detection. The framework combines IoT technology with DRL algorithms to adjust particle compensation methods in real time based on particle seeding distribution index (SDI) measurements taken across the river surface. The proposed algorithm is employed in newly developed PDCS that is mounted on the UAV, allowing for quick particle compensation in areas lacking sufficient features. Experimental results indicate that the proposed algorithm can quickly reduce the SDI values within the targeted regions of interest to the range of [0.75, 1], while successfully identifying and counteracting environmental disturbances. Field tests validate the system’s performance and reliability, offering an innovative approach to estimating river surface velocities in environments with insufficient natural flow features.

Ecological flow management is important for maintaining ecosystem stability and promoting sustainable development. Dynamic ecological flow regulation depends on precise real-time monitoring of water levels and flow velocities. To address challenges in ecological flow monitoring, including maintenance difficulties and insufficient accuracy, an improved DSC-YOLOv8n-seg model is proposed for dynamic multi-parameter sensing, achieving more efficient object detection and semantic segmentation. Compared with traditional affine transformation-edge detection, this approach enables joint recognition of water level lines and staff gauge characters, achieving an average recognition error of ±1.2 cm, with a model accuracy of 93.1%, recall rate of 94.5%, and mAP50:95 of 93.9%. A deep learning-based spectral principal direction recognition method was also employed to calculate the surface water flow velocity, which demonstrated stable and efficient performance, achieving a relative error of 0.005 m/s for the surface velocity. Experimental results confirm that it can effectively address issues such as environmental interference, exhibiting enhanced robustness in low-light and nighttime scenarios. The proposed method provides efficient and accurate identification for dynamic water level monitoring and for real-time detection of river surface flow velocities to improve ecological flow management.

Distinguishing anthropogenic perturbations from natural carbon cycling in mega-rivers remains a critical challenge. Carbon isotopes (δ13CDIC) provide a powerful tracer to decode these complex interactions which are often masked in traditional hydrochemical assessments. This study investigated the hydrochemistry and multiple stable isotopes (δD, δ18O and δ13CDIC) of the Yangtze River surface water (YRSW) during the dry season to quantify these contributions. δD values ranged from -117.8‰ to -44.6‰ and δ18O from -16.3‰ to -7.0‰, aligning with the Global Meteoric Water Line, which confirms atmospheric precipitation as the primary water source shaped by continental and altitude effects. DIC concentrations ranged from 1560 to 5724.29 µmol L-1 (mean 2993.79 µmol L-1), acting as a net CO2 source with an average pCO2 of 522.08 µatm. Stoichiometric and isotopic analyses reveal that carbonate weathering dominated by soil CO2 is the primary DIC source. However, in the middle and lower reaches, anthropogenic sulfuric/nitric acid weathering and organic matter oxidation were identified as key drivers decoupling DIC from natural climatic controls. This study systematically reveals for the first time the spatial differentiation patterns of DIC and δ13CDIC on the scale of the entire Yangtze River Basin and the main controlling factors, providing a new perspective for the study of carbon cycle in large river basins under high-intensity human activity interference.

Measuring river surface velocity is crucial for hydrological monitoring. However, traditional methods exhibit significant limitations under extreme weather conditions and emergency scenarios. In this study, we propose OF-PhyNet, an innovative hybrid neural network architecture designed as an image-based velocimetry method to improve the accuracy of river surface velocity estimation. By integrating an optical flow algorithm with a physics-informed neural network, OF-PhyNet achieves deep coupling of motion feature extraction with fluid dynamics constraints. To address the challenges posed by uneven particle distribution and illumination variations in natural rivers, we introduce a feature extraction strategy based on K-means clustering and morphological operations, which enables effective flow field segmentation and reliable identification of particle regions. Experimental validation conducted in the Heihe River Basin, using Stalker Pro II surface velocity radar (SVR) as a reference, demonstrates that OF-PhyNet achieves a mean absolute error of 0.054 m/s and a relative error of 7.22% across 12 cross-sectional measurement points, substantially outperforming PIVLab with 11.65% and the Horn-Schunck algorithm with 25.84% relative error. As a UAV-based image velocimetry approach, OF-PhyNet offers exceptional mobility for rapid flood monitoring and enables robust reconstruction of flow fields in particle-sparse regions, delivering precise non-contact river velocity measurements beyond the reach of traditional methods.

A machine learning-based algorithm for estimating river surface water velocity

Accurate estimation of river-surface flow velocity is critical for water-resources management, flood early-warning systems, and ecological modeling. Although unmanned aerial vehicle (UAV) image velocimetry offers high efficiency and noncontact operation, its robustness and accuracy degrade under complex conditions-such as illumination variability, low-texture surfaces, and nonrigid water-surface motion. In particular, detecting subtle surface displacements in low-texture regions remains challenging, thereby affecting flow-field reconstruction. To address these issues, we propose RSV-RAFT, a deep optical-flow framework for river-surface velocity assessment. First, we enlarge the receptive field of the feature-extraction module and introduce the Leaky ReLU activation to enhance sensitivity to weak textures and small-scale motion. Second, we design a Decoupled Enhanced Channel-Attention Flow Head (DECA-Flow Head) based on a spatial-channel decoupling architecture to improve modeling of minute displacements and boundary motion. Finally, we present a Deformable Coordinate-Aware Upsampling Mask Generator (DCAU-MaskGen) that integrates deformable convolutions with coordinate attention to enhance boundary perception and structural reconstruction. Experiments on multiple optical-flow benchmarks and real UAV river imagery show that RSV-RAFT outperforms existing methods in both accuracy and efficiency while maintaining a mean absolute percentage error (MAPE) of 10.5%-15.5%, demonstrating its potential for engineering applications.

Supervised deep learning methods have been widely employed to detect floating macroplastic litter (>5 mm) in (fresh)water bodies. However, few studies used them to quantify floating litter fluxes in rivers with wide cross-sections, that is important for pollution assessment. Additionally, commonly used supervised learning (SL) models rely on extensive labeled data, that is time-consuming and expensive to obtain. Moreover, regardless of the model type, current deep learning models for litter detection usually fail to correctly identify small litter items. To overcome these issues, we propose a semi-supervised learning (SSL)-based framework combined with Slicing Aided Hyper Inference (SAHI) for quantifying cross-sectional floating litter fluxes in rivers. The framework includes four steps: (a) collecting camera images of river surfaces from multiple locations across the river, (b) developing a robust litter detection model using SSL, (c) applying this model with SAHI to detect litter items in images, and (d) post-processing the detection results to quantify fluxes. The SSL method involves: (i) self-supervised pre-training of a ResNet50 on a large amount of unlabeled data, and (ii) supervised fine-tuning of a Faster R-CNN with the ResNet50 backbone on a limited amount of labeled data. We evaluated the in-domain detection performance of SSL models with varying pre-training epochs and pre-training dataset sizes, using images from waterways of The Netherlands, Indonesia and Vietnam, that were used for model pre-training and fine-tuning. Additionally, we assessed the zero-shot out-of-domain detection performance of SSL models and litter flux quantification performance of the proposed framework on a Vietnam case study, that was not used for model development. We benchmarked our results against the SL methods and human visual counting. The results show that SSL models benefit from longer pre-training time and larger pre-training dataset, achieving an in-domain F1-score increase of 0.2 and a zero-shot out-of-domain increase of up to 0.14, over baseline SL benchmarks. Furthermore, the SAHI method correctly identifies 45 additional small litter items (areas < 1,000 cm2), improving the F1-score by up to 0.19, compared to the results obtained without SAHI. The flux measurement results indicate that the SSL-based framework substantially underestimates fluxes by a factor of 3-4 compared to human measurements, due to missed detections of transparent litter items and items entrapped in water hyacinths. However, it estimates nearly twice the fluxes of the baseline SL-based framework, aligning more closely with human measurements. These findings highlight the potential of SSL-based framework to enhance litter flux measurement. Scaling it with broader datasets could significantly advance global-scale litter monitoring systems.