
ABSTRACT Dune maps are fundamental for the research on dune systems and a long‐standing objective in aeolian geomorphology. However, they are often generated manually and thus lack reproducibility and feasibility for extensive study areas. This study introduces Duneminds , a deep learning framework built around a U‐Net convolutional neural network for the automated detection and mapping of dune crests. The model was applied to the south‐western Kalahari linear dune field and tested with a range of input configurations, including different data sets, levels of data augmentation and resolutions. Data sets included Sentinel‐2 optical imagery, the Copernicus Digital Elevation Model and derivatives of the latter. Among these, models trained on optical imagery outperformed those based on elevation data, with the best‐performing one achieving an IoU score of 0.77. We applied the framework to generate a continuous crest map of the entire south‐western Kalahari dune field, reproducing dune crest pathways and patterns of the study area.
ABSTRACT Frequent surface geological hazards and ecological issues arise from coal mining beneath loess hills and gullies. The unique and complex surface geological conditions in loess plateau have made it challenging to analyse surface deformation patterns. To address this, this study focuses on Panel 9303 of a coal mine in Shanxi, employing drone aerial surveys, RTK field measurements, numerical simulation and theoretical analysis to investigate the differential surface deformation patterns of loess hills and gullies during shallow coal seam mining. The results indicate that during panel mining, the settlement curve exhibits a ‘concave’ shape over hills but a ‘convex’ shape over gullies. The maximum settlement reaches 8.21 m over hills, compared to 7.79 m over gullies. The magnitude and extent of horizontal movement along the mining direction are greater over hills than over gullies, while the horizontal movement perpendicular to the mining direction is significantly influenced by the downhill slope orientation. Higher gravitational loading due to slope inclination and compressive loading at the bedrock interface contribute to concentrated deformation over hills, rendering these areas prone to collapse fissures and step‐type fractures. The surrounding rock at the gully bottom experiences lower settlement due to compression from adjacent slopes. By developing a stress model for compression and antisettlement in gullies, the premining surface antisettlement condition equation T 1 cos γ 1 + T 2 cos γ 2 ≥ C s ‐ I s and the antisettlement coefficient s are derived. Substituting the geological conditions of gully mining into the equation confirms its feasibility. Combined with the antisettlement coefficient component s i , an evaluation method is proposed to assess the influence of various mining and geological parameters on gully settlement. Based on this evaluation method and surface failure characteristics, targeted prevention and control measures for mining and geological parameters during extraction, as well as postmining restoration techniques for surface damage, are studied. The findings provide theoretical foundations and guidance for geological hazard prevention and ecological restoration in mining areas.
Citizen scientists have long made important contributions to the monitoring and appraisal of environmental features and processes. However, they have tended to focus more on biological and water quality than on physical aspects of rivers. Nevertheless, in the last two decades, a variety of survey methods have emerged for recording physical processes, forms and human pressures and interventions. This has created a need for guidance on selecting appropriate methods and combining them effectively. Reflecting the many professional multi-scale physical frameworks that have been proposed since the 1980s, we present a simple, cost-effective framework for appraising the physical condition of river networks that can be applied by citizen scientists. The framework includes the range of spatial scales where citizen scientists can make significant additional contributions to information already routinely collected by river management agencies. We indicate the types of physical information that can be incorporated at the different spatial scales, and we provide examples of relevant survey techniques that are currently in use within the United Kingdom. We apply the framework to the River Crane, Greater London, to illustrate how it can be applied, using selected citizen science survey approaches, to enhance understanding of system functioning, answer specific questions and support river restoration site selection and monitoring. This case study illustrates the scale of work undertaken, the number of citizen scientists involved, the evolution of the survey programme, lessons learned and future intentions. We conclude by providing some broader perspectives on citizen river science and the physical framework approach.
ABSTRACT Assessing the spatiotemporal dynamics of large wood is critical for understanding the dynamic processes influencing river channels, floodplains and organisms using the riverscape. We quantified large wood in the lower White River, Colorado and Utah, which has an unregulated flow regime and has been invaded by woody vegetation, primarily Russian olive ( Elaeagnus angustifolia ). Using aerial photointerpretation, we quantified the spatial and temporal variation in large wood over 20 years (2004–2024) across 142 river km. To assess potential drivers of large wood dynamics, we used a generalized linear mixed effects model and quantified several geomorphic and hydrologic characteristics that we predicted would explain variation in wood density including channel width and slope, sinuosity, unit stream power, valley bottom slope, peak discharge and the standard deviation of channel slope and width. There was considerable variation in geomorphic and hydrologic variables among reaches and many variables were highly correlated ( r > 0.70). Wood densities varied significantly among reaches and increased throughout the study period in all reaches. We observed significantly more mid‐channel wood than bank‐attached wood in all reaches. Of the uncorrelated variables, sinuosity (beta = 0.38, SE = 0.09) and peak unit stream power (beta = 0.32, SE = 0.10) were both positively correlated with wood densities. Mean wetted width was negatively correlated with wood densities (beta = −0.29, SE = 0.06). Our results suggest heterogeneity in channel characteristics can create ‘sticky reaches’ where large wood densities are high and remain high over time. The consistent increase in wood densities across all reaches over time, despite declining flows, suggests that other variables, such as expansion of invasive woody vegetation, may be contributing to system‐wide increases in wood recruitment. Future studies should investigate how large wood dynamics might interact with changing channel morphology, changing riparian vegetation and decreasing flow to better understand the wood regime in dryland rivers.
ABSTRACT Geomorphic contrasts across drainage divides can reflect several interacting controls. We examine how contrasts across the shared Chalus–Karaj divide in the Central Alborz, Iran, vary among metrics and explicitly defined spatial populations. We combine basin morphometry, selected channel‐network diagnostics, fixed‐reference‐area topography, segmented Mχ as a relative steepness proxy and route‐based knickpoint associations with mapped faults and lithologic contacts. In the selected analytical networks, mean Mχ is higher in Chalus (184.35 vs. 129.73), and knickpoint densities are 2.903 versus 1.871 km −1 . At Aref = 1.0 km 2 , the primary 15‐endpoint population gives Chalus‐minus‐Karaj contrasts of +102.61 m for channel elevation, −0.0630 for mean upstream gradient and −43.02 m for R500. Four locally co‐supported segments cover 12.45% of the divide; their R500 contrasts range from −76.31 to +92.55 m. A separate raster‐based selection gives 19 catchments and a pooled R500 contrast of −26.26 m. For the same fitted Mχ field, primary upstream endpoint contrasts are +4.21 under equal‐node weighting and −4.88 under equal‐source weighting; a 17‐endpoint connector sensitivity gives +2.04 under equal‐source weighting. Route‐level fault and contact associations are modest and sensitivity dependent. These results support spatial and metric‐dependent geomorphic heterogeneity, with additional sensitivity to sampling population and averaging weights. They do not isolate an independent scale effect, a unique causal mechanism or a regional drainage‐divide migration direction or rate.