Earth and Environmental Sciences

Study region Yellow River Basin, China. Study focus While the impacts of climate change and human activities on river systems are widely recognized, decoupling their highly non-linear interactions in heavily engineered basins remains computationally and analytically challenging. This study addresses this gap by proposing an advanced hybrid deep learning framework that integrates multiscale statistical diagnostics, the Range of Variability Approach (RVA), a Particle Swarm Optimization-Long Short-Term Memory (PSO-LSTM) model, and Random Forest (RF) algorithms. This framework incorporates rigorous convergence validation for PSO optimization, multi-method cross-validation for change-point detection, and comprehensive collinearity diagnostics. This integrated approach is employed to systematically evaluate longitudinal hydrological alterations and precisely attribute the drivers of runoff evolution from 1960 to 2020 across the basin. New hydrological insights Across the Yellow River Basin, abrupt runoff reductions and the magnitude of hydrological alteration amplify consistently from headwaters to estuary, superimposed on a basin-wide 11–14 year climatic periodicity. A distinct spatial dichotomy governs driving mechanisms: climatic factors dominate runoff generation in the upstream source area, whereas cumulative anthropogenic interventions – including reservoir regulation and large-scale vegetation restoration – drive over 80% of the runoff decline in the middle and lower reaches. Widespread vegetation greening and land-use transitions have fundamentally shifted the basin’s blue-green water balance toward enhanced evapotranspiration, dampening peak flows and exacerbating downstream water scarcity. Upstream inflow acts as the absolute controlling factor for downstream hydrology; after removing its cascading effect, local climatic drivers remain dominant in the source area but contribute only marginally to downstream runoff. These findings clarify the full-basin longitudinal pattern of hydrological alteration and its spatially heterogeneous drivers, addressing a long-standing knowledge gap for the Yellow River Basin. They provide direct, actionable scientific support for differentiated water resource management and targeted ecological restoration planning across reaches, underpinning sustainable water security strategies for this water-scarce, highly engineered basin.

The Segment Anything Model 2 (SAM2) produces high-quality zero-shot segmentation of natural images, but two obstacles block its use at scale for remote sensing. First, the automatic mask generator trades coverage against quality: strict acceptance thresholds leave most of a scene unsegmented, while permissive thresholds admit imprecise masks. Second, large images must be tiled, which fragments objects across tile boundaries. This paper presents Remote SAMsing, an open-source pipeline that addresses both obstacles without modifying SAM2. A multi-pass algorithm re-runs SAM2 on each tile, painting accepted masks black between passes and relaxing thresholds only when coverage stagnates, so the most precise masks are captured first. A parameter-free best-match merge then reconstructs objects across tile boundaries. The pipeline requires neither training data nor manual threshold tuning, and tile size is its only scene-dependent setting. Across seven urban and agricultural scenes spanning 5 cm to 4.78 m ground sampling distance, coverage rises from 30%–68% with single-pass SAM2 to 91%–98%. The pipeline recovers discrete objects (buildings 95%, cars 82%–93% Det@0.5), with boundaries 3–8 × more precise than SLIC and Felzenszwalb. It transfers to MNF false-color imagery without retraining (99.5% ASA) and processes a 1.94-billion-pixel Potsdam mosaic at 97% coverage, with no loss of per-class quality. Remote SAMsing thus provides annotation-free segmentation of very large remote sensing images, ready for object-based image analysis and requiring no SAM2 expertise. • First pipeline to fully segment large RS images with SAM2 and no training. • Multi-pass with adaptive thresholds raises coverage from 30%–68% to 91%–98%. • Parameter-free boundary merge reconciles segments across tile boundaries. • No SAM2 expertise required: default config works across all tested scenes. • Tile size acts as implicit scale parameter for the detection trade-off.

Study region Study region: The upper reach of the Ganjiang River in Ganzhou, Jiangxi Province, China, where rapid urbanization and extensive levee systems have increased exposure to extreme flood hazards. Study focus A coupled 1D-2D hydrodynamic model (GAST) was applied to simulate an overtopping scenario (MY) and three controlled levee-breach scenarios (A, B, C) under 50-, 100-, and 500-year flood events. Breach discharge, downstream hydraulic responses, and spatiotemporal inundation were analysed to quantify how breach location and flood magnitude influence flood redistribution and downstream flood propagation. New hydrological insights for the region Controlled levee breaching modifies flood redistribution between the river channel and adjacent floodplains, with downstream hydraulic responses jointly regulated by breach location and flood magnitude. Upstream breaches produced the greatest downstream peak attenuation, reducing peak discharge by 3.87–10.31% and peak water level by 0.26–0.93 m through earlier redistribution of channel flow. In contrast, peak-delay mechanisms evolved with flood magnitude, as floodplain storage increasingly influenced flood-wave propagation during larger floods. Controlled levee breaching also reshaped the spatial distribution of flood hazards, increasing inundation area by 10.89–39.50% and inundation volume by 28.22–150.33% within receiving areas while reducing downstream flood impacts. These findings advance process-based understanding of flood redistribution in highly leveed urban rivers and provide transferable insights into downstream hydraulic responses in highly leveed river–floodplain systems with comparable channel–floodplain hydraulic controls.

Abstract Near‐surface wind speed (NSWS) regulates the intensity of land‐atmosphere fluxes and the viability of wind energy systems, playing a pivotal role in human adaptation to climate change. However, at the current stage, global climate models exhibit a wide inter‐model spread in NSWS simulations, limiting the reliability of future wind energy projections. Here, we identify the primary drivers of this uncertainty over China. We show that the spread is concentrated in North and Northeast China. Models simulating extensive forest loss exhibit weak or negligible NSWS trends because deforestation smooths the land surface and counteracts the background surface stilling. In contrast, models with limited forest loss or forest gain yield pronounced NSWS declines. This substantial spread is mainly attributable to the varying magnitude of forest cover changes, accounting for approximately 74% of the inter‐model spread. Our findings suggest that constraining forest cover changes is critical for substantially reducing the uncertainty in surface wind projections.

Abstract We investigate the nighttime equatorial ionosphere during the 19 January 2026 geomagnetic storm using Global‐scale Observations of the Limb and Disk OI 135.6 nm observations and ground‐based GPS Total Electron Content. The observations reveal a pronounced longitudinal asymmetry in the post‐sunset equatorial ionization anomaly (EIA). Over the South American sector, the two EIA crests converged and merged into a single equatorial peak across ∼35° longitude within ∼2 hr, while remaining separated with active equatorial plasma bubbles (EPBs) over the Atlantic and West African sectors. This behavior suggests a rapid, large‐scale post‐sunset crest merging under storm‐time conditions. Concurrent radar and magnetometer measurements indicate storm‐time westward electric fields drove the large‐scale merging. The timing of these electric fields relative to local sunset suppressed prereversal enhancement and inhibited EPB growth west of ∼45°W. These electrodynamic changes are consistent with a brief southward followed by prolonged northward interplanetary magnetic field Bz, suggesting overshielding‐driven downward drifts.

What makes long-term policy issues rise to the legislative agenda in the state-level policymaking process? More specifically, why have sea level rise adaptation policies gained traction in some states but not others? The following paper presents a case study of the State of Hawai’i, which is critically vulnerable to sea level rise impacts. The study operationalizes the concept of policy traction or “issue salience” as the number of issue-related bills submitted to the state legislature in a given annual legislative session and finds a dramatic increase in the number of issue-related bills proposed in the Hawai’i State Legislature starting in 2018. What caused this abrupt rise in issue salience around the topic of sea level rise? The paper employes Kingdon’s classic multiple streams framework to analyze a convergence of three “streams” of activities (the problem recognition stream, the policy proposal stream, and the political stream) that created a “policy window” for the issue to gain traction. The study finds that, while all three streams contributed to opening the policy window, the problem recognition stream in particular played a key role in driving the uptick in sea level rise-related policy after 2018. For the problem recognition stream, the creation of the state-wide Sea Level Rise Exposure Area tool and accompanying Vulnerability Assessment, released in 2017, provided crucial data that allowed policymakers to define the scope of sea level rise impacts and to devise more technically feasible solutions.

Abstract Hydrothermal systems along the Mid‐Atlantic Ridge (MAR) are spatially variable sources of dissolved organic carbon (DOC) to the deep ocean, with implications for carbon cycling and trace‐metal transport. We present DOC measurements from four research cruises focusing on major vent fields at Menez Gwen, Rainbow, Lost City, Broken Spur and TAG (25–40°N). Distinct DOC patterns were observed across sites. At Rainbow and Menez Gwen, DOC maxima (>1,000 μM) were decoupled from Fe and 3 He plumes, indicating diffuse DOC sources and potential chemosynthetic production. In contrast, CTD casts intercepting buoyant plumes at TAG and Rainbow showed elevated DOC (90–140 μM) coincident with Fe and 3 He, consistent with direct hydrothermal inputs and possible Fe–ligand complexation. No DOC enrichment was detected at Broken Spur. These results demonstrate that hydrothermal vents have highly variable magnitudes and sources of DOC, leading to spatially heterogeneous impacts on deep‐ocean carbon cycling and metal‐organic interactions.

Study region The Dahaize Coal Mine lies on the southern margin of the Mu Us Sandy Land, a key part of the large coal energy base in northern Shaanxi, China. Study focus Deep coal mining severely disturbs regional groundwater stability, yet the long-term and multi-scale responses of hydrogeological functions of multi-layer aquifers in mining areas remain poorly understood. This study combines hydrochemical approaches with multiple isotopic tracers (δ 2 H, δ 18 O, δ 34 S, δ 18 O sulfate , and 222 Rn) to elucidate spatiotemporal evolution features and dominant driving mechanisms of multi-layer aquifer hydrogeological functions across the whole coal mining cycle. New hydrological insights for the region Phreatic water (PW) and Anding Formation confined water (ADCW) maintained stable hydrogeological functions. PW was mainly affected by sulfate leached from coal gangue under rainfall infiltration, while ADCW was slightly improved owing to weakened vertical leakage recharge. By contrast, Zhiluo Formation confined water (ZLCW) suffered severe functional degradation driven by expanding water-conducting fracture zones (WCFZs). WCFZs lowered hydraulic gradients, slowed groundwater flow and promoted evaporite dissolution. As preferential flow pathways and geochemical reaction interfaces, WCFZs strengthened hydraulic connectivity between ZLCW and mine water (MW), accelerated cation exchange, carbonate dissolution and radon emission, and caused simultaneous enrichment of SO 4 2 - , δ 18 O, δ 34 S, and 222 Rn in the two aquifers. This study deepens the understanding of the driving mechanisms for the hydrogeological functional evolution of aquifers in mining areas.

Abstract We applied an integrated UAV‐ and ground‐based magnetometry workflow to map shallow subsurface fractures and cavities in the town of Grindavík during the 2023–2025 volcano‐tectonic crisis. Short‐wavelength linear and semi‐circular magnetic lows delineate open or partially open fractures and localized cavities, including features with no surface expression, verified through field observations, LiDAR surface deformation, and targeted excavations. Several anomalies correspond to reactivated fractures identified through historical aerial imagery. Magnetic forward modeling shows anomaly shapes and amplitudes compatible with fracture and void sources within the upper ∼10–20 m. The approach enabled rapid mapping of major fracture networks in inaccessible terrain, revealing fracture continuity, aperture variability, and branching patterns that guided hazard assessment and emergency response. This study demonstrates the first combined UAV‐ and ground‐based magnetometry application for near real‐time fracture mapping during an urban volcano‐tectonic crisis with implications for rapid hazard response in other tectonically active settings with sufficient magnetic contrast.

Abstract The identification and distribution analysis of salt marsh vegetation in Yangtze Estuary are essential for conservation, management and ecological sustainable development of coastal wetlands in this region. However, the accuracy of identification and analysis depends on the model performance and the number of collected sample points used for training. Therefore, this study develops a Bayesian-optimized Random Forest(BoRF) classification workflow for salt marsh vegetation mapping and combines it with an Unmanned Aerial Vehicle(UAV)- assisted sample collection strategy. After obtaining the optimal feature combination through the comparative analysis, the vegetation classification and precision evaluation were carried out using multi-temporal Sentinel-2A imagery and the sample points collected by a DJI Mavic 3M drone, in four representative areas of Yangtze Estuary: Chongming East Beach, Jiu Duan Sha, Nanhui East Beach, and Nanhui Coastal Beach. The experiment results indicate that, among the four compared methods, BoRF achieves the best overall performance, with an overall accuracy of 0.88 and a Kappa coefficient of 0.85, and reduces misclassification among vegetation types with similar spectral characteristics. In addition, the UAVassisted sample construction strategy improves classification accuracy by approximately 3% compared with conventional field-survey-based sampling. These results demonstrate the practical value of the proposed workflow for salt marsh vegetation classification in complex coastal wetland environments.

Groundwater systems are shaped by deeply interwoven physical processes and social dynamics, yet traditional hydrogeological approaches have often treated these domains separately. Over the past decade, socio-hydrogeology has emerged in response, offering an approach that systematically incorporates social dimensions into hydrogeological investigations. In its initial formulation, socio-hydrogeology was meant to raise awareness among hydrogeologists of the social implications of their work, fostering engagement with water users and interdisciplinary collaboration from the onset of research. Since then, the approach has gained academic traction and institutional recognition, yet it still lacks a clear conceptual definition, posing risks to its coherence and impact. This paper marks 10 years since the term’s inception and critically reflects on the development and future direction of socio-hydrogeology. By critically analysing the evolution of the concept and situating it among broader inter- and transdisciplinary frameworks, this paper proposes an updated conceptualisation to clarify its disciplinary scope, practical relevance, and further encourage its application. Attention is also given to the implications for hydrogeological education, arguing for a shift in pedagogy that better equips researchers to engage with the ethical and cultural dimensions of groundwater science. Socio-hydrogeology offers an opportunity to reframe hydrogeology not only as a natural science but as one deeply shaped by (and accountable to) society.

Cities face rising heat exposure due to climate change, yet the extent to which urban heat islands (UHI) intensify relative to overall warming remains unclear. This study examines Singapore, a highly urbanised tropical coastal city, focusing on near-surface air temperature (Tas) and UHI intensity patterns under future climate and land-use scenarios using a 100 m urban modelling system (uSINGV). We analyse diurnal cycles and spatial distributions to understand the influence of sea breezes and assess cooling effects from one million additional trees. Key findings include: (1) although Tas increases significantly in future climate, island-wide UHI intensity remains comparable to historical periods during the day and decreases slightly at night; (2) island-wide UHI intensity exhibits a distinct diurnal pattern, positive at night driven by heat storage and reduced ventilation but negative during the day due to cooler marine air advected to urban areas by sea breezes; (3) one million additional trees provides modest spatial cooling without substantially altering island-wide Tas or UHI compared to the dominant influences of sea breezes and background warming. These results suggest that in coastal cities, mesoscale ventilation primarily controls daytime urban heat patterns and can locally suppress or reverse UHI despite rising temperatures.

Abstract This paper presents the first results of a coordinated experiment on the ionospheric plasma response to high‐frequency (HF) overdense heating, conducted using the SURA heating facility and the Ionosfera‐M satellite. The experiment combined ground‐based observations with in situ satellite measurements, enabling a comprehensive study of artificial ionospheric disturbances. During the experiment, the SURA facility transmitted HF waves into the ionosphere toward the magnetic zenith, while Ionosfera‐M and ground‐based instruments measured various ionospheric plasma characteristics in the modified region. The penetration of the pump wave energy up to the satellite height was observed as well as some broadening of VLF noise spectrum and signatures similar to the main ionospheric trough in the HF‐disturbed region. The data analysis revealed a coincidence of the regions of artificial airglow generation and electron density depletion near the magnetic zenith. The results demonstrate an efficiency of combining ground‐based and satellite studies of the HF‐pump induced phenomena.

Water plays a crucial role in the transport properties of silicate liquids and is especially important in subduction-zone magma source regions where conditions promote complete miscibility between water and silicate melts. However, accurately tracing the movement of fluids and volatile-rich magmas is prevented by poor knowledge of the variation of viscosity across silicate- H 2 O joins at relevant pressure-temperature conditions. We used molecular dynamics driven by a machine learning potential of ab initio quality to investigate supercritical NaAlSi 3 O 8 − H 2 O fluids from 0 - 100 % H 2 O at 1200–3000 K (at 2 GPa) and 1–5 GPa (at 2500 K). Our results provide atomic-level insight into how H 2 O disrupts the silicate melt network to change viscosity. We find smaller fluid viscosity at a given pressure, temperature, and H 2 O content than obtained in previous models. We fit our results to an empirical viscosity model that allows application to transport in geologic settings. For model subduction zones, we find that the mean percolation velocity of fluids ascending by Darcian flow varies strongly with H 2 O content and that, at constant H 2 O , the fluids accelerate as they rise from slab source into the mantle wedge, and decelerate on further ascent. For plausible water contents, we find that ascent times from the slab to the Moho may be as little as 8000 years, satisfying constraints from U-Th disequilibria.

Abstract Post‐sunset equatorial plasma bubble (EPB) formation is believe to be influenced by factors including thermospheric neutral winds. This study characterizes pre‐sunset neutral winds observed by ICON on EPB occurrences observed by GOLD over the east Atlantic. 3 years of data during the months of November through February are utilized when most EPBs are seen in this region enabling statistical analysis of the data. This is accomplished by generating two populations to compare: winds that precede EPB occurrence and those which do not. Both populations' wind speed magnitudes, wind speed gradients with local time, and wind speed gradients with altitude can then be compared. Zonal and meridional winds weakly correlate to EPB occurrence and are therefore not strong predictors of EPB occurrence. Strongest correlation occurs 1 hr before sunset for meridional winds. Sample sizes are insufficient to determine if pairs of gradient samples were statistically distinct.

Abstract The spatial distribution, sources and ecological risk of 16 US EPA priority polycyclic aromatic hydrocarbons (PAHs) in surface soil from Liuxin area, Xuzhou, China was investigated in this study. Gas chromatography-mass spectrometry (GC-MS) analysis revealed ΣPAH concentrations ranging from 115.66 to 373.53 ng∙g-1 (mean: 196.26 ng∙g-1), with phenanthrene (Phe) exhibiting the highest mean concentration (51.18 ng∙g-1). Spatial analysis demonstrated a southeast-to-northwest concentration gradient, with maximum PAH levels observed in southeastern zones. PAH composition showed dominance of low-molecular-weight species (2,3 rings: 46.7 %), followed by 5,6 ring PAHs (28.3 %), and 4-ring compounds (24.9 %). Source apportionment using molecular diagnostic ratios (DR) and positive matrix factorization (PMF) identified four primary contributors: coal-fired power generation & metallurgical emissions (34%), biomass combustion (31%), diesel fuel combustion/leakage (29%), traffic emissions (6%). Ecological risk assessment employing Dutch soil quality standards revealed localized exceedances in southeastern areas, particularly for fluoranthene (Fla) and Phe. However, toxic equivalency quotients (TEQs) remained low, with incremental lifetime cancer risks (ILCRs) below 10-6 for all exposure pathways. These findings indicate that while PAH contamination warrants monitoring, current levels pose negligible carcinogenic risk to the local population.

Dolomitization is a key diagenetic process that reorganizes porosity and permeability in carbonate rocks, yet the coupling between interface kinetics and transport heterogeneity remains poorly constrained. We performed time-resolved hydrothermal dolomitization (metasomatism) experiments on ooid grainstone and chalk. We integrate to our results new datasets from published experiment on marble. Hence, we investigated how carbonate rocks and crystalline metamorphic textures impact the dynamics of replacement fronts. Pseudo 4D micro-CT, along with SEM, EBSD and microprobe analyses reveal that the mechanism of dolomitization depends on the texture of the host-rock and the chemistry of the diagenetic fluid. The three investigated rocks have different grain size, porosity and permeability. In both grainstone and chalk, the resulted dolomite rims propagate rapidly and produce wavy reaction fronts, invoking a progressive replacement inward within the host-rock. Applied mass-balance calculations across the observed replacement interface highlight a positive volume change between the grainstone and marble, and a negative volume change regime in the chalk. Texture, i.e. the combined effect of grain and pore size distribution, crystallographic properties and chemical content, controls the regime of the metasomatism. Roughness characterization of the reaction front shows that the front propagates randomly (H = 0.5) under the swelling regime observed in the marble and grainstone, while it shows a persistent behaviour in the shrinking regime of the chalk (H = 0.8). Beyond offering a framework in which established replacement processes are linked together as depending on texture, this work paves the way for using roughness analysis of replacement fronts as a different diagnostic tool to understand the mechanisms at work during metasomatism.