Earth and Environmental Sciences

Abstract. Ultra–low-noise inertial sensors are a cornerstone of modern geoscientific instrumentation, enabling high-resolution observations across seismology, geodesy, gravimetry, and vibration isolation. Achieving and reliably predicting their performance requires a rigorous treatment of physical causality, noise propagation, and uncertainty, particularly in force-feedback architectures operating near fundamental limits. In this study, we introduce a causal and uncertainty-aware digital-twin framework for the design and metrological assessment of ultra–low-noise geoscientific inertial sensors. The proposed framework integrates mechanical dynamics, force-feedback control, transduction, and digital acquisition within a physically realisable model that explicitly enforces causality and stability constraints. Starting from a minimal equation-of-motion description, the digital twin is formulated in the frequency domain to construct causal transfer functions and a comprehensive noise-budget model. The framework enables the systematic separation of fundamental thermal noise limits from implementation-dependent noise sources, including readout, actuation, and digital acquisition effects. We introduce quantitative performance metrics based on self-noise spectra, dominant noise regimes, crossover frequencies, and near-plateau bandwidths, allowing complex spectral behaviour to be condensed into actionable design indicators. Parameter uncertainties are propagated through the digital twin to provide uncertainty-aware performance estimates and robustness diagnostics. Through a series of illustrative analyses, we demonstrate how the proposed digital twin supports informed design trade-offs, identifies performance bottlenecks, and prevents non-physical or overly optimistic sensitivity estimates arising from non-causal modelling assumptions. While focused on inertial sensors, the methodology is general and transferable to other classes of geoscientific instruments. The framework provides a transparent and extensible foundation for next-generation sensor design, virtual experimentation, and metrologically consistent performance prediction.

Estuaries are vital ecosystems that support biodiversity, human populations, and economic activities such as fisheries, aquaculture, and tourism. Managing estuarine systems often involves interventions to regulate sediment and nutrient inputs with the goal of improving water quality, and subsequently the economic values of the estuary. However, these management strategies can produce unintended consequences. The Chesapeake Bay is an estuary recovering from eutrophication which provides clear case studies for these patterns. In the Chesapeake Bay, decreased sediment inputs, resulting from shoreline stabilization and watershed modifications, can impair marsh resilience to sea level rise and disrupt sediment-light-algal interactions affecting water clarity. Likewise, reductions in riverine nutrient inputs, while beneficial for mitigating eutrophication, may contribute to surface water acidification. These outcomes complicate assessments of restoration success. Throughout other estuaries’ recovery from historical nutrient and sediment pollution, similar patterns have been observed in several other estuarine systems worldwide. Effective long-term estuarine management requires integrating diverse monitoring approaches, adaptive decision-making, and ecosystem-based strategies to balance environmental and economic objectives. Understanding both the intended and unintended effects of management actions is crucial to ensuring sustainable outcomes along the non-linear path of recovery from eutrophication, including both nutrient and sediment pollution. Future efforts should prioritize holistic assessments, stakeholder involvement, and targeted modeling to guide effective estuarine conservation and restoration.

Moderate-resolution leaf area index (LAI) retrieval over heterogeneous landscapes is affected not only by unresolved subpixel composition in coarse-resolution predictors, but also by structural bias in supervisory labels aggregated from higher-resolution products. To address this issue, we developed a reference-guided two-stage workflow to improve LAI retrieval from the Visible Infrared Imaging Radiometer Suite (VIIRS). In the first stage, aggregated Sentinel-2 LAI was calibrated against Ground-Based Observations for Validation (GBOV) LP3 reference LAI using subpixel plant functional type (PFT) fractions and forest-sensitive hinge terms to generate corrected 500 m labels. In the second stage, a random-forest model was trained using VIIRS spectral reflectance, viewing geometry, vegetation indices, texture, and subpixel compositional variables. Model development was based on 2020–2021 data from 11 U.S. GBOV sites. Performance was evaluated by same-site temporal transfer to 2019 and 2022 and by strict leave-one-site-out (LOSO) validation. Label calibration improved agreement with GBOV from a coefficient of determination (R2) of 0.752 and a root mean square error (RMSE) of 1.110 to an R2 of 0.908 and an RMSE of 0.676. Under LOSO validation, the final model achieved an R2 of 0.901 with an RMSE of 0.703. On the 2019/2022 overlap subset shared by the final VIIRS retrieval, the official VNP product, and the GBOV reference, the final model achieved an R2 of 0.905 and an RMSE of 0.609, compared with 0.755 and 0.978 for the official VNP product. These results show that reference-guided label correction, combined with explicit subpixel compositional information, can substantially improve VIIRS LAI retrieval over mixed pixels within the evaluated study domain.

Landslides are among the most common and destructive geological hazards and pose a significant threat to the long-term stability of infrastructure systems. In particular, long-distance power transmission corridors often traverse mountainous and forested regions, where landslides can endanger tower foundations and transmission line safety. Such landslides predominantly occur in sloped forested areas, where dense vegetation causes severe occlusion that blurs landslide boundaries and creates strong visual similarity with surrounding land covers. Consequently, accurate and efficient landslide identification from remote sensing imagery remains a significant challenge. To address these challenges, we propose a structural constrained contrastive learning network (SC-Net) for reliable landslide extraction from remote sensing images. First, a multi-structural feature extraction module is designed to capture landslide-specific geometric characteristics. These features are further enhanced by fusing multi-scale semantic representations extracted from a pretrained backbone network through an attention-based adaptive feature fusion module. Additionally, a mask-constrained object-level contrastive learning strategy is introduced to enforce global structural consistency at the landslide object-level, thereby improving the discriminability between landslide and non-landslide regions. Extensive experiments conducted on the publicly available CAS landslide dataset demonstrate the effectiveness of the proposed method. The proposed SC-Net achieves IoU scores of 89.89% and 79.76% on the CAS-UAV and CAS-SAT datasets, respectively, outperforming the best-performing baseline by 2.09% and 0.46%. The proposed method provides an effective solution for large-scale landslide monitoring and demonstrates potential for applications in power transmission corridor inspection and infrastructure safety assessment.

Ground-level ozone (O3) has become a major air pollutant in China following PM2.5, particularly in the southeastern coastal region, where the frequent interaction of typhoons and the subtropical high complicates pollution control. In this paper, spatial autocorrelation and a multiscale geographically weighted regression (MGWR) model were employed to estimate the spatiotemporal heterogeneity and driving mechanisms of O3 in the Southeast Coastal urban agglomerations from 2015 to 2024. Temporally, the annual average O3 concentration exhibited a fluctuating trend of an initial increase, followed by a decrease and a subsequent rebound. A bimodal monthly pattern was observed, with peaks in May–June and August–September and minima in winter. Diurnally, the concentration showed a consistent pattern of being higher in the daytime and lower at night, peaking in the afternoon, driven by solar radiation and temperature. Spatially, O3 exhibited a distinct north–south gradient, with the highest in Jiangsu Province, followed by Shanghai, Zhejiang and Guangdong, and the lowest in Fujian. Significant spatial autocorrelation was detected, with hot spots in the Yangtze River Delta and cold spots in Fujian and adjacent areas. Seasonally, the most severe pollution with the greatest spatial heterogeneity, occurred in summer, contrasting with the uniformly low concentrations in winter. Compared with OLS and GWR, the MGWR demonstrated superior explanatory power. O3 was jointly influenced by precursors, natural factors, and socioeconomic factors, with the influence intensity ranked as follows: NO2 > average elevation > population density > annual precipitation> wind speed > built-up area > proportion of the secondary industry in GDP. Notably, the effects of NO2, annual precipitation, and the proportion of the secondary industry exhibited strong spatial heterogeneity, operating at finer spatial scales. These findings provide scientific support for sustainable air quality management and region-specific O3 control in southeastern coastal China.

Abstract Lake water surface elevation (WSE) data in Sweden are needed to understand water supply and ecosystem services provision. SWOT can complement WSE in situ measurements, but the influence of lake shape on its accuracy remains unclear. We explore the performance of the SWOT pixel‐cloud and this influence across 62 lakes from August 2023 to November 2025. We found that SWOT estimated WSE with a mean error of about 5 cm in lakes larger than 1 km 2 . Winter ice reduced the number of usable observations in northern lakes, but SWOT still captured major WSE changes when data were available. The influence of lake morphology on SWOT accuracy occurred mainly through shoreline effects reflected through the compactness of the lakes. These results show that SWOT provides useful WSE information in Sweden, but that lake morphology cannot be ignored.

In this study, we explored how early bark beetle attacks can be detected using weekly aggregated Sentinel-2 data in combination with static data, such as geo- and forestry data. We used an XGBoost classifier, known for its strength and reliability in classification, and compared three sets of data: static data only, satellite data only, and using both together. Having trained the models on cumulative weekly data, we were able to track changes in model performance and feature importance over time, identifying key weeks for the detection of bark beetle attacks. A systematic overview of feature importance identified the Red-edge 3 and blue Sentinel-2 bands as the most important when combined with static data; it also showed changes in feature importance compared to using satellite-only data, e.g., adding static features reduced the importance of red and red-edge 2 bands. Among the static features, land cover and landforms were the most important. Evaluating the temporal features for the combined model highlighted certain weeks as containing key information for detection: week 19, which was the main swarming week; week 25, which is 6 weeks after swarming and just before the second generation is completed; and weeks 31 and 33, more than 3 months after the tree was attacked and well after the new generation has swarmed. The study shows that combining static features with cumulative Sentinel-2, accumulated across weeks, are all important for improving the detection of bark beetle attacks, and that such ideas form an important part of early warning systems.

Forest Fires alter vegetation structure, photosynthetic activity, and regeneration dynamics. However, the thermal effects on nearby unburned vegetation remain poorly understood. Heat radiation generated during fires spreads in all directions and may cause dehydration, physiological stress, and foliar damage in surrounding plant communities. In the Pantanal, one of the largest wetlands on Earth and an ecosystem naturally poorly adapted to fire, these indirect impacts have received little scientific attention. Therefore, this study aimed to quantify and delineate the spatial gradient of the indirect impacts of Forest Fires on adjacent unburned vegetation, using NDVI as an indicator of changes in photosynthetic vigour between 1985 and 2022. A database integrating vegetation indices, burned areas, and land use and land cover information was assembled. Burn scars mapped by MapBiomas were surrounded by concentric buffer zones ranging from 50 to 250 metres to evaluate NDVI variation before and after fire events. Kruskal-Wallis and Dunn tests combined with K-means clustering were applied to identify spatial and ecological response patterns among vegetation types. Significant differences (p < 0.05) were detected up to 100 metres beyond burn scars, indicating measurable indirect fire effects. Forest and wetlands showed greater resilience while grassland formations exhibited slower recovery overall.

Remote sensing image semantic segmentation plays a pivotal role in converting complex image data into quantifiable geographic spatial information, underpinning applications such as disaster assessment, urban planning, and agricultural resource investigation. Fully supervised semantic segmentation methods rely heavily on labour-intensive pixel-level annotations, prompting a shift towards weakly supervised semantic segmentation (WSSS) that utilizes image-level annotations. However, remote sensing images are characterized by dense, intricate targets and the absence of distinct backgrounds, leading to challenges, such as sparse activation of local regions, incomplete localization in class activation maps (CAMs), and noisy, rough boundaries in pseudo-labels generated from image-level supervision. To address these issues, we propose a deep learning method with pixel relationship constraints for WSSS in remote sensing images. Specifically, we design an image reconstruction (IR) loss function to provide pixel-level supervision, enhancing the completeness of CAMs; a pixel relationship constraint (PRC) module to strengthen the global correlation of target regions and improve detailed information extraction; and an intersection optimization strategy (IOS) based on the Segment Anything Model (SAM) to refine pseudo-labels and segmentation results by mitigating noise. Here, we show that our method achieves mean Intersection over Union (mIoU) values of 63.85%, 71.20%, and 40.96% on the Vaihingen, Potsdam, and iSAID datasets, respectively, reaching 89.25%, 89.02%, and 65.66% of the performance of fully supervised methods. This work advances WSSS for remote sensing images by addressing key limitations of CAM-based pseudo-labels generation, offering a cost-effective alternative to fully supervised approaches and facilitating broader applications in geographic information science and earth observation. The code is available at https://github.com/CHENDL-SHEN/PRCCAM.

Fluid injection into the subsurface can induce seismicity by reactivating shear rupture, which typically produces larger earthquake magnitudes than tensile rupture. In laboratory shear rupture experiments, pressurization of the entire fault is often limited because large unconfined samples allow fluid to leak at free surfaces. In this study, we investigated shear fault reactivation by directly injecting fluid into a PMMA fault (760 mm long, 76 mm high) formed as the interface between two separate PMMA blocks. To prevent leakage in the 76 mm dimension, we made a low permeability barrier by coating the outer edges of the fault with Teflon tape. Fluid pressure then extended along the 760 mm dimension, resulting in seismicity migration away from the injection well. Changes in injection rate and fluid viscosity revealed two mechanisms: (1) Low injection rate or low-viscosity fluid caused seismicity migration governed by pressure diffusion, and (2) high injection rate or high-viscosity fluid caused seismicity migration proportional to injected volume. Simulations with a 2D poroelastic model showed that seismicity migrated with the fluid pressure front in the volume-driven regime, whereas fluid pressure advanced well ahead of seismicity in the pressure-diffusion-driven regime. These results highlight that Teflon tape effectively sealed faults and controlled fluid flow, and that injection rate and fluid viscosity have a strong impact on fault slip and induced seismicity.

Crop distribution mapping is critical for safeguarding national food security and optimizing agricultural management strategies. However, accurately distinguishing spectrally similar crops while maintaining precise boundaries remains a significant challenge. To address this, we propose an enhanced High-Resolution Network, termed HRNet-ICoT-CNN. The key innovation of this model is the Improved Contextual Transformer (ICoT) module, which dynamically integrates local structural details with global temporal contexts, thereby enhancing multi-scale spatiotemporal feature extraction without compromising spatial resolution. Applied to Sentinel-2 time-series NDVI data from Xinhe County in the Aksu region of Xinjiang, China, the model accurately delineated the distributions of key crops – cotton, wheat, and maize – and was evaluated against leading existing models. Experimental results indicate that HRNet-ICoT-CNN outperforms baseline models across all major evaluation metrics, achieving an overall accuracy (OA) of 94.91%, average accuracy (AA) of 81.81%, and Intersection over Union (IoU) values of 0.8507 for cotton, 0.6467 for wheat, and 0.5513 for maize. Generalizability tests further affirmed the model’s robustness, yielding an OA of 81.47% and an AA of 73.20% in two untrained counties, with IoUs of 0.8750, 0.9065, and 0.6314 for the three crops, respectively. These results highlight the model’s strong transferability and reliability, offering a robust scientific foundation for crop mapping via remote sensing in arid regions.

Abstract Natural and man-made cellulosic textile fibres and microfibres have been viewed as environmentally benign due to their biological origins. This assumption has caused natural and man-made cellulosic microfibres to be overlooked and underrepresented in research despite findings demonstrating their abundance in environmental systems, compared to synthetic, microplastic fibres. This systematic literature review critically examines how natural and man-made cellulosic fibres undergo multiple chemically intensive processing stages within the textile manufacturing processes, resulting in materials that differ substantially from their raw counterparts, exposing critical gaps in understanding across environmental science, toxicology and textile production. These chemical processes influence fibre degradation, increase the potential for pollution adsorption, and contribute to the environmental persistence and impact of microfibre shedding. Across studies, natural and man-made cellulosic microfibres are frequently reported in high abundance in environmental systems yet remain underrepresented in environmental monitoring which predominantly focus on synthetic microplastics.&amp;#xD;By adopting a textiles perspective, which highlights viewing information through a textile-focussed lens rather than from an environmental science perspective which has dominated this field. This review challenges the assumption that natural and man-made cellulosic microfibres pose no harm to the environment, suggesting that “natural” fibres can cause comparable harm to the environment compared to their synthetic, microplastic counterparts. This demands a reframing of current narratives that position natural and man-made cellulosic textiles as environmentally benign, recognising their potential for environmental harm. The findings further support the inclusion of chemically processed natural and man-made cellulosic microfibres in environmental monitoring, policy and sustainable textile design.

Accurate hyperspectral image classification is fundamental to geospatial applications but is often constrained by annotation scarcity. To achieve high classification performance under small-sample conditions, we propose the Memory-Guided Adaptive Spectral–Spatial Perception model, which incorporates a three-level globalization strategy. At the single-sample level, an adaptive perception Transformer combines deformable and dilated convolutions with a Transformer encoder to capture global context from individual samples. At the intra-batch level, we introduce a metric learning strategy that explicitly captures structural dependencies and feature relationships among samples within each mini-batch, enabling comprehensive feature aggregation in a localized context. At the cross-batch level, a memory-guided strategy constructs a dynamic memory bank to store and retrieve features from same-class samples across training iterations, bridging past and present distributions to enhance generalization. Using only 1% of the SaliLMSS, Pavia University and Kennedy Space Center datasets and 0.5% of the WHU-LongKou dataset as training samples, our method achieves outstanding overall accuracy of 96.15%, 97.81%, 89.22% and 99.32%, respectively, outperforming existing methods.

Anthropogenic CO2 emissions have perturbed the global carbon cycle and increased atmospheric carbon concentrations to critical levels, making carbon capture and storage (CCS) a key strategy for mitigating climate warming. Natural carbon sequestration has operated continuously in marine environments throughout Earth history. Here, we investigate the growth mechanisms and carbon-sink significance of calcite concretions in the Mesoproterozoic Xiamaling Formation from the Zhaojiashan section and the Zhenzhuquan section in the North China Craton, using petrographic, elemental geochemical and C-O-Re-Os isotopic evidence. The presence of erosional surfaces and local truncation of host-rock laminae suggests that these concretions formed synsedimentarily or during early diagenesis near the sediment-water interface. The δ13C values (−5.05‰ to 1.54‰) of samples, together with δ18O-δ13C relationships, indicate a marine carbonate affinity and suggest that dissolved inorganic carbon was the dominant carbon source. In addition, the concretions display initial 187Os/188Os ratios as low as 0.136, close to the mantle Os end-member, implying a contribution from mantle-derived material during concretion formation. The middle rare earth element and yttrium (MREYs)-enriched patterns and slight positive Ce anomalies further indicate that concretion growth occurred mainly within the Mn- and Fe-reduction zones. We estimate that the calcite-concretion-bearing interval of the Xiamaling Formation sequestered 70.24 Gt C, equivalent to 257.56 Gt CO2, serving as an archive of marine carbon burial in the Mesoproterozoic ocean. Microbially mediated carbonate precipitation may represent an effective carbon immobilization mechanism in marine sediments and has potential implications for the development of subseafloor carbon storage strategies, especially where biocatalysts and/or brine could accelerate seawater CO2 mineral trapping to industrially relevant rates.

Conventional remote sensing faces challenges in regions characterized by complex terrain, frequent cloud and fog cover, image scarcity, and mixed pixels. This study develops an integrated framework that sequentially combines spatiotemporal fusion, water masking, feature engineering, and machine learning to enable long-term SSC retrieval. Key results include: (1) RASTFM achieved the highest spectral fidelity among four fusion algorithms; (2) development of an adaptive machine learning model for independent hydrological stations and a water body masking strategy; (3) mainstream SSC decreased by 21.2% temporally and a 93.8% longitudinally from 2006 to 2020, generating 15-year along-channel mainstream SSC data for the Three Gorges Reservoir area.

Traditional methods for measuring the soil-water characteristic curve (SWCC) are time‑consuming and prone to disturbing soil structure, whereas resistivity testing offers rapid, non‑destructive advantages. However, its applicability to compacted silty clays under drying‑wetting cycles has not been systematically validated. In this study, three types of silty clay (HZ, WY, ZJ) used for dam construction were compacted to degrees ranging from 92% to 100%. Under both drying and wetting paths, resistivity (two‑electrode method), volumetric water content, and matric suction (filter paper method) were measured simultaneously. A resistivity synthesis parameter (Re), derived from Archie’s structural factor and the shape factor, was introduced to establish an empirical relationship between matric suction and Re. The results show that the Pearson correlation coefficients between resistivity and matric suction range from 0.895 to 0.979 (p &amp;lt; 0.01), and partial correlation analysis confirms that water content is the dominant controlling factor. At the same water content, resistivity during the drying path is consistently higher than that during the wetting path, exhibiting clear path dependency. A higher compaction degree reduces the rate at which resistivity decreases with increasing water content. The quadratic model fits the h‑Re relationship well, with R² &amp;gt; 0.95 for all soils under both paths. The predicted matric suctions are in close agreement with those fitted by the classical van Genuchten model, with most data points lying near the 1:1 line. These findings confirm that resistivity parameters can effectively capture the hydraulic hysteresis of silty clays. The proposed empirical h‑Re equation provides a rapid, non‑destructive approach for estimating matric suction under reservoir water level fluctuations and holds practical value for seepage stability monitoring and early warning of earth‑rock dams.

Archaeological feature detection at complex, mixed-environment sites requires accurate, efficient methods for identifying subtle morphological signatures. This study presents a semi-automated remote sensing pipeline for the detection and delineation of archaeological earthworks at Cahokia Mounds (Illinois, USA), a major Mississippian urban center and UNESCO World Heritage Site. Three LiDAR datasets, two collected via UAV-mounted sensors and one from a piloted aircraft survey, were processed into Digital Terrain Models and transformed into Local Relief Models (LRM). K-means clustering was applied to segment the LRMs into feature classes, followed by contour bounding using the OpenCV library to outline mounds and borrow pits. Additionally, SAR-derived Local Incidence Angle (LIA) rasters from PALSAR-3 and Sentinel-1 were processed through angular deviation mapping to identify slope anomalies associated with archaeological features. Results across all five datasets demonstrate the complementary strengths of LiDAR and SAR: LiDAR excels at resolving elevation-defined features such as mound footprints, while LIA captures directional slope behavior that highlights mound edges, borrow pit rims, and linear features such as causeways. Comparative analysis of LiDAR acquisition frequencies reveals minimal differences in archaeological feature recovery between pulse settings, suggesting that sensor platform choice matters more than power-density tradeoffs for this application. Despite the need for human review to filter modern disturbances and natural false positives, the integrated workflow meaningfully accelerates prospection and reduces interpretive subjectivity. The methods are scalable, site-invariant, and work with open-access data, making them applicable to archaeological landscapes worldwide.

Abstract Enhanced geothermal systems rely on increasing permeability and pore surface area in rock. Cyclic thermal shocking can achieve both by inducing thermal cracks through repeated rapid cooling. Laboratory experiments subjected micritic limestone, granodiorite, and trachybasalt to up to 10 thermal shock cycles, while tracking crack evolution qualitatively using time‐lapse electron microscopy and quantifying pressure‐dependent permeability and elastic wave velocities. This work advances prior efforts focused primarily on crack initiation by demonstrating how lithology‐specific microstructures govern the cyclic evolution, persistence, and efficiency of pressure‐dependent permeability enhancement during cyclic thermal shocking. This reframes microstructure as a key design variable controlling permeability enhancement and monitoring during geothermal stimulation. Contrasting mineral thermal properties, large mineral grains, and irregular vugs promote the greatest permeability enhancement. Velocity reductions were most pronounced &lt;10 MPa effective pressure ( P eff ) and diminished with increasing cycle number, indicating that velocity‐based monitoring in geothermal systems must account for P eff and cycle number.

3D Gaussian Splatting (3DGS) offers efficient explicit rendering, but large-scale remote-sensing use remains fragmented across UAV/aerial photogrammetry, satellite reconstruction, large-scene scaling, surface modeling, and geospatial evaluation. We present a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-informed scoping review based on 55 core studies identified through Web of Science, Scopus, IEEE Xplore, and supplementary searches completed on 3 June 2026. A faceted taxonomy organizes the literature by platform, sensor model, scalability strategy, and geometric supervision. The synthesis shows that partitioning, hierarchy, compression, and feed-forward inference improve scalability but do not guarantee metric geometry. Reliable deployment additionally requires sensor-consistent projection, geometric or georeferencing constraints, explicit supervision labels, and product-level evaluation. In control-point-free settings, internal consistency should be distinguished from independently validated accuracy. We therefore propose a platform-aware benchmark framework that jointly records visual fidelity, computational cost, metric geometry, product utility, failure behavior, and reproducibility metadata for UAV/aerial, satellite, and hybrid settings.

Hyperspectral unmixing aims to extract pure endmembers and their corresponding abundance from mixed pixels. Existing deep learning-based unmixing methods predominantly rely on convolutional neural networks (CNNs) or Transformer architectures. However, CNNs suffer from limited receptive fields and struggle to capture long-range spectral dependencies across the entire spectral sequence. While Transformers possess global modeling capabilities, they are constrained by quadratic computational complexity and lack the ability to adaptively filter redundant noise in consecutive spectral bands. To address these limitations, this paper proposes a dynamic hyperspectral unmixing network integrating a spectral sequence Mamba with local spatial–spectral awareness. Specifically, the network features a novel asymmetric dual-stream collaborative architecture. The first branch, the spectral sequence Mamba, models hyperspectral data as a one-dimensional continuous sequence and employs the selective state space model to perform global scanning with linear complexity. This adaptively filters redundant spectral bands to accurately extract high-purity global spectral semantics. The second branch, dedicated to local spatial–spectral awareness, uses an attention-augmented CNN to capture local continuous spectral variations and spatial textures, providing fine-grained geometric boundary constraints for abundance estimation. Furthermore, a spatially adaptive gated fusion module is designed to dynamically balance global spectral semantics and local spatial–spectral details according to the pixel mixing complexity of varying spatial regions. Extensive experiments on multiple public hyperspectral datasets demonstrate that the proposed method achieves significant improvements in unmixing accuracy over comparative methods.

Continental shelves along tropical semi-arid margins remain sparsely mapped at high resolution, limiting process-based interpretations of bedforms dynamics and reducing quality of baseline information required for marine spatial planning and offshore development. In this study we present an integrated geomorphological and sedimentological baseline for two middle- to outer-shelf sectors of the Ceará Basin, on Northeast Brazil. We combined multibeam echosounder bathymetry and backscatter, side-scan sonar imagery, and sediment sampling to map bedform morphology and sediment textures in detail across more than 12 km 2 in total. High-resolution mapping reveals strong spatial contrasts in seabed morphology and acoustic texture across the two areas. The Western Area comprises a heterogeneous mosaic of geomorphic elements, including a channel, clustered circular depressions, and a terrace-like bench, associated with marked variability in backscatter and grain size, from granule-grade sediment on the bench to sandier substrates in depressions and channel-floor domains. The Eastern Area is dominated by sand-wave and organized into a crest–trough system, with geomorphometric patterns and side-scan sonar textures pointing to spatial variability in seabed roughness and bedform expression. Together, these results highlight how inherited shelf morphology and present-day hydrodynamic reworking interact on a sediment-limited shelf to produce contrasting bedform–sediment domains over short along-shelf distances. The mapped bedform-sediment domains provide a robust physical baseline to guide future process studies and to inform monitoring and risk-aware planning for infrastructure and marine spatial management on the Brazilian semi-arid continental shelf.

Analyzing the impervious surface area (ISA) of China’s three major urban agglomerations and expansion pattern of representative cities is of great significance for formulating context-specific urban planning policies and enhancing the resilience of urban development. This research employs impervious surface expansion indicators to reveal the spatiotemporal dynamics of impervious surface area, and further applies the location centrality index (LCI) to investigate the expansion structures and patterns of five megacities. The results indicate that the ISA of China’s three major urban agglomerations expanded from 22,544.10 km2 to 69,348.59 km2 from 1985 to 2024; both the expansion rate and intensity have exhibited a distinct decelerating trend. However, significant regional variation exists in the peak expansion phases: the Guangdong–Hong Kong–Macao Greater Bay Area Urban Agglomeration (GBA) peaked during the Initial Exploratory Phase (1985–1995), whereas the Yangtze River Delta Urban Agglomeration (YRD) and Beijing–Tianjin–Hebei Urban Agglomeration (BTH) reached their peaks during the High-speed Development Phase (2005–2015). At the urban scale, LCI-based analysis of urban expansion across five megacities reveals that adjacency expansion type has become the predominant form of growth, exceeding 50% in most cities. These findings enhance our understanding of evolutionary trajectories of ISA within urban agglomerations and the expansion patterns of five megacities, providing a scientific basis for sustainable urban planning and urban resilience.