Earth and Environmental Sciences

Changes in urban form strongly affect surface thermal conditions, yet long-term quantitative assessments of this relationship, particularly the role of ventilation resistance, remain limited. To address this gap, this study integrates XGBoost, SHapley Additive explanations (SHAP), and multi-scale geographically weighted regression (MGWR) to examine how six morphological, ecological, and human-activity factors influence land surface temperature (LST) in Kaifeng City. The results indicate three main findings. First, LST increased significantly from 1986 to 2024, while interannual variability declined, indicating a gradual reduction in regional thermal fluctuations. Second, NTL was consistently the dominant indicator across the five representative years, while BF and NTL together captured the effects of urban expansion and intensified human activity. Third, FAD coefficients were more spatially heterogeneous in urban fringe areas than in the urban core. In 2020, the dispersion of FAD coefficients in fringe areas was 2.74 times greater than that in the central area, indicating stronger spatial differentiation in ventilation-related morphological constraints during urban expansion. Although FAD made only a modest contribution to overall predictive accuracy, it provided supplementary diagnostic information not captured by conventional density indicators and showed nonlinear, directional, and spatially heterogeneous responses. Compared with previous studies that mainly examined short-term or single-dimensional relationships between urban morphology and LST, this study integrates building densification, ventilation-related morphological resistance, ecological conditions, and human activity intensity into a long-term LST-driver framework, providing evidence to support heat-risk management during urban regeneration and outward expansion.

Accurate estimation of rice chlorophyll status is essential for precision nitrogen management. Although multispectral and hyperspectral sensors are effective, their high cost and data-processing complexity limit widespread adoption. This study investigated the efficacy of low-cost UAV-based RGB imagery for estimating rice chlorophyll status, represented by SPAD measurements, focusing on the effects of image resolution, feature integration of vegetation indices (VI) and texture features (TF), and TF parameterization. UAV RGB imagery with three image resolutions (3.8, 9.6, and 25 mm) was analysed, and six regression algorithms, including multiple linear regression (MLR), support vector machine (SVM), random forest (RF), Gaussian process (GP), light gradient boosting machine (LGBM), and categorical boosting (CAT), were used to establish estimation models. Results demonstrated that 1) the integration of VIs and TFs improved model performance compared with VIs alone, and the best results were obtained at the 25 mm resolution; 2) TF parameterization also affected model performance, and the best results were obtained with a window size of 11 × 11 and a direction of 90°. Validation on an independent test dataset confirmed the model’s reliability (R2 = 0.84, RMSE = 2.35). Overall, these findings provide a high-efficiency, cost-effective framework for rice chlorophyll status monitoring and RGB-based decision making in precision nutrient management.

Using accurate land cover data is essential to monitor land use and cover changes and assess the effectiveness of various environmental policies. This study evaluates the accuracy of contemporary global land cover products with 10 m spatial resolution, including Google’s Dynamic World (GDW), European Space Agency’s World Cover (ESA WC) and Esri Land Cover (ELC) in mapping forested areas in Poland, aiming to test an assumption if the combination of these products may improve forest mapping accuracy compared to any individual product. Three global datasets and their combinations were assessed with the 2022 EU Land Use/Cover Area Frame Survey (LUCAS). A land cover map of Poland (S2GLC PL) for 2021 served as an auxiliary reference data set. Forest cover classification accuracy was evaluated using precision, recall, and F1-score metrics, and spatial agreement of binary forest maps in the thematic global products was measured with the Intersection over Union (IoU) at two various scale levels (country and province). Our results showed that forest mapping accuracy of three global products varies for Poland, with F1-score equal to 72.2% for ELC, 76.9% for ESA WC, and 68.8% for GDW. IoU against S2GLC PL was equal to 82.6%, 82.3% and 75.2%, for ELC, ESA WC and GDW, respectively, and slightly exceeded 70.5% for three global products. A specific combination of binary forest maps from global products, where the output forest area consisted of forests mapped at the same time by all three products and forests mapped at the same time only by GDW and ESA WC yielded better accuracy indicators than any single product and other tested combinations (F1-score equal to 80.4%, and IoU against S2GLC PL equal to 87.1%).

Accurately assessing the spatiotemporal evolution of ecological environment quality (EEQ) on the Loess Plateau of Northern Shaanxi is of great significance for consolidating the ecological security barrier of the Yellow River Basin. Most of the existing research focuses on a single ecological theme, which does not reflect the overall ecological status of the region. In this study, a remote sensing ecological index (RSEI) model was constructed to systematically assess the EEQ from 2000 to 2024. The Theil–Sen estimator, Mann–Kendall test, and Hurst exponent were jointly employed to detect change significance and predict future trends, while the Geodetector model was applied to explore driving factors. The results were as follows: (1) EEQ exhibited a fluctuating but overall upward trend, with the mean RSEI rising from 0.376 in 2000 to 0.545 in 2024—an average annual increase of approximately 0.00569. (2) Spatially, a distinct pattern of “higher in the south, lower in the north and the lowest in the northwest” was observed. Over the 25-year period, the combined proportion of “excellent” and “good” grades increased by roughly 20 percentage points, and the “moderate” grade expanded from 13.61% to 47.12%. (3) Areas showing an improving trend accounted for 91.21% of the total area and highly overlapped with those projected to improve in the future. (4) Single-factor detection revealed that geomorphological type exerted the greatest influence on the spatial heterogeneity of EEQ, with a multi-year mean q-value of 0.701. Interaction detection further indicates that the geomorphology–land use interaction may continue to shape the regional EEQ’s spatial distribution. These findings provide a scientific basis for precise ecological restoration planning and spatial optimization on the Loess Plateau of Northern Shaanxi.

Accurately estimating village-level winter wheat yield in coastal saline–alkali farmland is challenging because this region has strong spatial differences and multiple environmental stresses. In this study, Huanghua City, Hebei Province, was selected as a typical coastal saline–alkali area. Sentinel-2 images, climate factors, and topographic variables, including elevation, topographic wetness index, distance to the coastline, and distance to water systems, were combined to build a phenology-guided feature set for winter wheat yield prediction in coastal areas. The results showed that Phenology-Guided Feature Integration XGBoost achieved an R2 of 0.6382 and an RMSE of 450.15 kg/ha, which was slightly better than Gradient Boosting (R2 = 0.6256) and Random Forest (R2 = 0.6098), and clearly better than SVR (R2 = 0.4792), Ridge regression (R2 = 0.4582), and a single Decision Tree (R2 = 0.3088). Then, a three-stage branch was designed to identify the main drivers of SI, NDVI, and winter wheat yield at different stages, helping explain how environmental constraints and vegetation responses jointly affect final yield. The Three-Stage Fusion XGBoost Model achieved an R2 of 0.6439, an RMSE of 446.24 kg/ha, and an MAE of 363.38 kg/ha, showing a slight improvement in prediction accuracy. SHAP analysis showed that SI, distance-related factors, elevation, TWI, and NDVI were important drivers of winter wheat yield variation. Spatial prediction results showed higher winter wheat yield in inland areas (5145 kg/ha) and lower yield in coastal areas (4198 kg/ha). This framework supports village-scale winter wheat yield prediction in coastal saline–alkali farmland and improves model interpretability.

Abstract. The article includes an overview of the current state of knowledge regarding climate in Poland (Central Europe) in the 16th century and its changes. For this purpose, we utilised all previously published reconstructions and five new quantitative reconstructions incorporating dendrochronological data and documentary evidence. New dendrochronological data were used to reconstruct the mean winter or late winter–early spring temperatures, while documentary evidence enabled the reconstruction of mean winter (DJF) and summer (JJA) temperatures. The climate of Poland in the 16th century, as reconstructed from documentary evidence, was colder than it is today (1991–2020), particularly in winter (by 3.6 °C). In summer, it was only 0.7 °C colder than today. Compared to the average for the entire 20th century, however, the summer average in the 16th century was 0.3 °C warmer, whereas the winter average was 2.5 °C colder. In both dendrochronological reconstructions of the temperature of south-eastern Poland, the temperatures in the 16th century were generally lower than those recorded today (1951–2000), particularly in the case of the reconstruction based on the fir chronology (December–March). Anomalies, however, both positive and negative, were usually of less than one standard deviation from the long-term mean. On the other hand, in northern Poland, the February–March temperatures in the 16th century were, on average, comparable to those of the present. Most available temperature reconstructions for Poland reveal cooling over the last few decades of the 16th century, particularly during the winter half-year. The climate in the 16th century was more continental than it is today.

Currently, existing point cloud semantic segmentation methods do not fully exploit surface geometric features. In particular, the depiction of object boundaries and the transition areas of curved surfaces is rather rough. On the other hand, the neighbourhood aggregation mostly follows a single strategy, making it difficult to simultaneously take into account the context and fine-grained differences and ignoring local details. To address these issues, this paper proposes a geometry-enhanced adaptive local feature aggregation network (GALA-Net). First, a geometric information embedding (GIE) module is introduced, which extracts pseudo-normal vectors and pseudo-curvatures of local point cloud regions as geometric priors, and incorporates multi-frequency sine–cosine encoding to capture multi-scale spatial relationships, yielding enhanced local geometric representations. Then, an adaptive feature fusion (AFF) module dynamically allocates fusion weights between semantic and geometric features, thereby alleviating channel coupling and neighbourhood noise amplification caused by simple concatenation. Next, a dual-path adaptive attention aggregation (DAAA) module jointly models semantic and positional attention and adaptively fuses them with max-pooled features to improve the robustness of local aggregation. In addition, a self-enhanced attention encoding (SEAE) module is designed to expand the feature representation space by extracting features through independent mapping branches and fusing them in a residual manner. The proposed model is evaluated on the S3DIS and ScanNetV2 datasets, achieving mIoU scores of 78.0% and 71.6%, respectively, which demonstrates its strong segmentation performance on indoor scenes.

Estimating chlorophyll-a (Chl-a) concentrations from satellite ocean color data remains challenging in the Arctic, where freshwater inputs, colored dissolved organic matter (CDOM), suspended particles, and low sun elevation alter optical properties and influence blue–green reflectance. Here, we combine satellite and in situ observations to examine how freshwater-driven optical variability shapes satellite-derived Chl-a across the Canadian Arctic Archipelago (CAA). Continuous underway observations were collected by a FerryBox system aboard the MS Roald Amundsen during August–September 2022 and matched with MODIS-OC3M Level-3 Chl-a (4 km, ± 2 days, 0.1° bins; n = 758). Satellite-derived Chl-a showed large differences relative to in situ observations, with a mean positive bias of 0.69 log 10 units and a root-mean-square error of 0.73 log 10 units, corresponding to an approximate 4.9-fold difference. These differences were strongly structured by environmental gradients, with the largest discrepancies occurring in low-salinity, CDOM-rich waters influenced by the Mackenzie River and decreasing eastward toward clearer, marine-dominated regions of Lancaster Sound. Previously-developed Arctic-tuned algorithms were applied to examine how regional models represent these gradients with the CAA. These approaches reduced overall bias and also resulted in substantial spatial variability linked to freshwater and optical gradients. To further account for these nonlinear environmental effects, a generalized additive model (GAM) incorporating salinity, CDOM, and temperature was applied, resulting in closer agreement between satellite-derived and in situ Chl-a, particularly in the Kitikmeot Sea. These findings demonstrate that freshwater-driven optical variability is a primary control on the calculation of satellite-derived Chl-a in Arctic shelf systems and that integrating environmental predictors into observational frameworks improves the interpretation of ocean color data in optically complex regions.

Multimodal fusion unlocks the potential of point cloud semantic segmentation, thereby driving advancements in surface observation and visual perception tasks. Although light detection and ranging (LiDAR) systems capture precise 3D structural geometry and optical images provide rich semantic and textural information, existing fusion methods struggle with limited cross-modal perception and insufficient information complementarity. To address these limitations, we propose a multi-stage LiDAR-image collaborative perception fusion network (MCPFNet) for point cloud semantic segmentation of urban scenes. At the middle fusion stage, the network incorporates an elevation-guided geometric-aware fusion module and a semantic-aware cross-attention fusion module to enable bidirectional feature injection between LiDAR and image modalities. In the late fusion stage, a bidirectional adaptive fusion module further refines semantic representations through gated weighting and bidirectional cross-attention mechanisms. Extensive experiments on three multimodal datasets with different resolutions, i.e., ISPRS Vaihingen, N3C-California, and UAVScenes, demonstrate that MCPFNet outperforms existing fusion methods, achieving mIoUs of 74.51%, 95.15%, and 62.76%, respectively. Hence, our multi-stage fusion and bidirectional interaction strategy is more reliable and accurate than existing methods in performing segmentation across diverse and complex urban scenes.

Accurate soil salinity estimation under small-sample agricultural conditions continues to pose a formidable challenge, attributed to the scarcity of labeled data, inherent representational limitations of single-backbone neural networks, and the heightened complexity of subsurface salinity inversion. To mitigate these intertwined challenges, this study developed a UAV-enabled soil salinity estimation framework that integrated lightweight convolutional neural networks and staged feature optimization, leveraging both RGB and multispectral imagery. A feature selection framework integrating random forest recursive feature elimination (RF-RFE), the one-standard-error (One-SE) criterion, and variance inflation factor (VIF) analysis was employed to reduce 129 candidate variables to a unified 16-channel feature set, which served as the common input for estimating both surface and subsurface soil salinity. Three lightweight single-backbone (VGGNet, ResNet, and DenseNet) and dual-backbone feature-level fusion networks (DenseResNet, DenseVGGNet, and ResVGGNet) were constructed and systematically evaluated for their performance in estimating both surface and subsurface soil salinity. Among the single-backbone networks, ResNet yielded the highest overall statistical accuracy, while DenseNet exhibited superior performance in preserving estimation trends. For surface soil salinity estimation, ResVGGNet achieved the best performance among all evaluated models, with an R2 of 0.820, RMSE of 0.626 g/kg, MAE of 0.409 g/kg, and RPD of 2.31 on the test dataset. SHAP analysis further highlighted the dominant role of vegetation and salinity-sensitive indices, together with selected spectral mean features, and revealed spatially complementary response patterns among major input channels. Collectively, the integration of lightweight multi-backbone feature-level fusion with streamlined feature optimization strategies effectively enhances the accuracy, robustness, and interpretability of UAV-enabled soil salinity estimation, particularly under the constraint of small agricultural sample sizes.

Subtropical mountain economic tree plantations are constrained by pronounced spatial heterogeneity in resource availability, yet the spatial scales at which soil properties, topography, and canopy structure regulate vegetation vigor remain poorly resolved. To address this gap, a spatially consistent multi-scale dataset combining 10 m high-density soil sampling, UAV-LiDAR, and multispectral remote sensing was used to quantify the scale-dependent drivers of the Leaf Chlorophyll Index (LCI) across 3–50 m within a Chinese hickory (Carya cathayensis Sarg.) plantation. The relative contributions of canopy, soil, and topography to LCI were decomposed across scales using an interpretable machine-learning framework (XGBoost–SHAP). At fine scales (3–10 m), vegetation vigor was primarily controlled by tree-level canopy structure, particularly tree height, reflecting localized resource acquisition. At intermediate scales (10–20 m), a distinct coupling window emerged, characterized by increased interaction complexity: LCI was predominantly driven by interactions between canopy structure and soil nutrient availability, whereas single-factor effects weakened. Notably, at 20 m this interaction pattern largely weakened and reverted to single-metric dominance. At broader scales (>30 m), complex interactions re-emerged, and dominant SHAP contributions shifted from nutrients and canopy structure toward topography and soil texture. These findings reconcile strong fine-scale drivers with weaker predictability at intermediate extents and demonstrate that soil–canopy relationships reorganize across spatial scales rather than remaining static. On the basis of these findings, a scale-hierarchical framework for precision forestry is proposed that aligns management interventions with the ecological scales at which dominant correlates operate across spatial supports.

Spaceborne Synthetic Aperture Radar (SAR) provides all-weather, day and night and wide-area imaging capability, and plays a critical role in maritime surveillance. While substantial progress has been achieved in SAR ship detection, SAR ship classification remains relatively underexplored, mainly due to the scarcity of reliable category labels. Automatic Identification System (AIS) provides vessel identity, type, and dynamic trajectory information, and thus offers vessel type information that is difficult to obtain directly from SAR imagery. This paper proposes a fine-grained nine-category SAR ship classification method based on AIS association, which reorganizes the original AIS vessel types into nine fine-grained categories of SAR ship, transfers AIS vessel type information to SAR detection through a global optimal matching strategy, and supports SAR-only vessel category recognition. By retaining only high-confidence SAR and AIS matched pairs and cropping the corresponding SAR ship chips, an SAR ship classification dataset containing 4472 ship chips across the nine categories is constructed. In Monte Carlo experiments based on real AIS records, the proposed association strategy achieves more reliable high-confidence label generation than the compared association methods under close ship ambiguity, spatial perturbation, distractor AIS candidates, and AIS static size errors. In the benchmark experiment on the constructed classification dataset, ConvNeXt-Tiny achieves the best performance among the compared mainstream classifiers. These results demonstrate that AIS association can provide reliable category supervision for SAR ship classification, and the trained classifier can perform ship classification using SAR imagery alone.

Imagery is fundamental to modern scientific research, making robust intrinsic camera calibration indispensable for accurate visual inference. The checkerboard-based calibration method has long been favored for its simplicity and ease of deployment and is widely used even in mission-critical computer vision pipelines. However, its limitations in modeling high-precision camera geometry can compromise downstream performance in tasks requiring geometric accuracy. In this work, camera calibration is revisited through the lens of photogrammetric self-calibration (PSC), and it is demonstrated that the PSC consistently outperforms the checkerboard method in both accuracy and precision across a range of vision tasks, including 3D reconstruction with structure from motion (SfM), visual simultaneous localization and mapping (SLAM), and novel-view synthesis and reconstruction. Our findings advocate for a paradigm shift toward calibration methods that better reflect the physical and projective properties of camera systems in real-world deployments for critical computer vision applications.

Marine spatial planning (MSP) is gaining prominence as an ocean governance tool in the Pacific Islands, yet its application and transformative potential remain under-examined. Drawing on participant observations and interviews with Pacific actors, this study critically assesses whether MSP is fit for purpose in advancing transformative ocean governance in the Pacific Islands. An integrative framework linking collaborative, polycentric, and adaptive governance to six IPBES transformative change clusters is used to evaluate the results. Findings indicate that MSP in the Pacific Islands currently operates in a reformist mode, contributing to improved coordination, dialogue, and awareness of ocean-use conflicts rather than deep governance transformation. Collaborative governance is evident in the rise of interministerial ocean committees and co-creation processes, yet meaningful engagement of Indigenous Peoples and Local Communities in decision-making remains weak. Polycentric governance is partially visible through multi-actor steering committees but often manifests as fragmented and weakly coordinated, with limited decision-making authority. Adaptive governance is the least developed dimension, constrained by gaps in data, technical expertise, institutional continuity, and learning mechanisms. Yet, emerging MSP practices, including knowledge co-creation and revitalization of community rights, indicate nascent shifts toward transformative ocean governance. The study concludes that MSP in the Pacific Islands contexts examined in this study holds transformative potential but will require deeper institutional restructuring, authority redistribution, sustained knowledge co-production, and stronger adaptive learning systems to disrupt business-as-usual governance.

This special issue addresses emerging technologies and future directions in air quality research by integrating spaceborne observations with in situ measurements, data fusion frameworks, and advanced computational techniques. The collective findings of the contributing studies offer a valuable resource for researchers, practitioners, and policymakers seeking to understand and quantify air pollution across diverse environments. The methodologies presented across these papers establish a foundation for identifying pollution sources and characterizing pollutant transport and transformation processes at the regional scale, supporting the stabilization of air quality management systems. To reduce the health burden of ambient air pollution, the contributing authors collectively underscore the need to raise awareness around reducing anthropogenic emissions and to advance space-driven data fusion systems, including expanding monitoring infrastructure, operationalizing AI/ML-driven analytical pipelines, implementing science-informed emission control policies, and fostering meaningful community engagement.

Metal sulfides are promising for capturing gaseous Hg 0, but their adsorption capacities are limited due to Hg 0 adsorption only on superficial unsaturated sulfur (S) coordination. Therefore, enabling saturated S coordination beneath the surface to participate in Hg 0 adsorption is an innovative and effective strategy. In this study, surface adsorption of Hg 0 on CuS was converted into a replacement reaction with a stoichiometric ratio of 1:6 through intermittent acid washing, allowing 2/3 of sulfur in CuS to contribute to Hg 0 adsorption, significantly increasing the capacity theoretically to 348 mg g –1 . The formation of a Cu vacancy due to the dissociation of four-coordinated Cu1 with one dangling bond and the transformation of four-coordinated Cu0 without a dangling bond to Cu1 alternately happened during each acid washing and Hg 0 adsorption, respectively, resulting in a chain reaction of the substitution of Cu in CuS by Hg 0 . Ultimately, 1/4 of four-coordinated Cu 1.5+ in CuS can be isomerous-substituted by Hg and dissociated into the acid washing solution, resulting in the growth of one-dimensional HgS in destroyed CuS crystals. CuS with intermittent acid washing demonstrates significant potential for recovering gaseous Hg 0 from flue gas, with Hg content in spent CuS exceeding 20%, offering a valuable approach for resource recycling.

A cross-disciplinary literature on just transitions shows that shifting to a lower-carbon economy can take jobs, tax revenues, and people from fossil-fuel-dependent communities. But few studies examine how pollution left by extractive industries may amplify and perpetuate losses. We conducted interviews with 39 stakeholders involved in remediating waterways polluted by abandoned coal mines in Pennsylvania, which once led coal production in the U.S. We find that, in our cases, water pollution acts as a “signal of neglect” that suppresses recreational use and economic activity and fosters pessimism among local residents. In our sample, communities with more social capital, defined by strong leadership and well-organized fishing and environmental networks, have successfully remediated numerous waterways. Although there was not deliberate coordination between remediation and economic development initiatives, participants report that cleaned waterways support emerging tourism and outdoor recreation industries. Our findings suggest that in some contexts, reversing signals of neglect can foster community optimism and local economic resilience. However, current policies to support waterway remediation depend on existing social capital and may therefore bypass socially-frayed communities. Our findings motivate policies to build grassroots capacity and to streamline remediation processes to reduce burdens on that capacity.

Forage fish are small, schooling fish that play a vital role in marine ecosystems, serving as a primary food source for seabirds, marine mammals, and commercially important species such as salmon. Despite their ecological importance, forage fish distributions remain understudied in the Canadian Salish Sea. Improved knowledge of their habitat uses and population dynamics is essential for effective conservation and management. In this study, a predictive geospatial model was developed to identify and map forage fish species “hot-spots” in the Canadian Salish sea. The model was constructed using 2,629 observations of six common forage fish species taken from Fisheries and Oceans Canada’s juvenile Pacific salmon survey, spanning 2000-2023. Discussions with a panel of local experts were then conducted to identify 18 environmental variables that were likely to influence the distributions of the forage species. Forage fish occurrence probabilities were estimated using a stacked ensemble approach which combined Neural Network, Generalized Linear Mixed Effects, and XGBoost models. The stacked ensemble achieved an average AUC of 0.73 across a 3-fold cross-validation, indicating strong overall ability to distinguish between presence and absence. Predicted presences were accurate 98% of the time (precision), and the model successfully detected 72% of all true presences (recall). Wind-driven surface current speeds, dissolved organic nitrogen (DON), and mesozooplankton biomass were found to be the most important predictor variables. A predictive map of forage fish hotspots was generated and reviewed, providing insight into the spatial distribution of the six species. Hotspots were identified around the Fraser River delta, along the southern coast of Vancouver Island, and throughout the Gulf Islands. Generally, higher probabilities were associated with inlets, such as Desolation Sound. The lowest probabilities of occurrence were associated with deep waters in the central Strait of Georgia.

To address the challenges of severe optical attenuation and dynamic feature extraction for moving target depth retrieval in complex underwater remote sensing environments, this paper proposes a dynamic target depth estimation method based on multi-source data fusion. Taking optical RGB imagery and neuromorphic vision (NeuroIV) data as joint inputs, the proposed method constructs a three-channel feature extraction and fusion network. By leveraging a hypergraph structure, it establishes association weights between dynamic (temporal) and static (spatial) nodes to capture spatiotemporal correlations. To efficiently process the high-dimensional multi-modal data, the traditional dot-product attention is replaced with element-wise multiplication, significantly reducing computational complexity. Furthermore, a lightweight deformable attention pyramid (DAP) and diffusion model is introduced to refine depth image edges, effectively suppressing discontinuities and abruptness in the estimation results. Compared to single-modality optical imagery, the fused multi-modal data yields a superior signal-to-noise ratio and foreground contrast, achieving an improvement of over 20% in the MAE index. These results validate the effectiveness and superiority of the proposed multi-modal fusion strategy for dynamic target observation and depth retrieval in aquatic environments.

Remote sensing ship detection plays a pivotal role in maritime surveillance, safety assurance, and traffic management. However, current detection methods often face significant challenges due to complex sea-surface background noise, large target-scale variations, and edge-hardware limitations. In this paper, we propose RepLite-YOLO, a lightweight detection framework based on YOLOv11n. Specifically, to alleviate irreversible spatial information loss during downsampling, we adopt the ADown module, originally introduced in YOLOv9, to generate spatially complementary features through its two-branch downsampling mechanism. This design helps preserve salient hull-edge responses while suppressing part of the random sea-surface interference, thereby improving feature robustness for small ship targets. To achieve substantial structural streamlining while maintaining competitive representational capacity under strict hardware constraints, we design the C3k2_OREPA_RS module, utilizing online re-parameterization (OREPA) to efficiently reconstruct deep layers without additional re-parameterization-induced inference operations. Furthermore, we construct the ELANFusion_Block by integrating Depthwise Separable Convolutions (DSC) into the ELAN paradigm to alleviate the multi-scale aggregation bottleneck, and tailor the Detect_DWLite head for highly compressed decoupled prediction. Experimental results show that RepLite-YOLO achieves a favorable balance between detection accuracy and computational efficiency. Compared with YOLOv11n, it reduces the number of parameters by 57.4% and GFLOPs by 49.2%, while maintaining competitive detection accuracy with slight mAP@50 improvements of 1.2 and 1.3 percentage points on the Vessel dataset and Ship Detection dataset, respectively.

Object detection in UAV aerial imagery is challenged by dense small targets, large-scale variation, complex backgrounds, and strict onboard computation limits. To address these issues, this paper proposes SMMNet (Structured-diffusion Mamba Mixture Network), a lightweight plug-and-play detection framework evaluated with YOLO family detectors. SMMNet contains three modules. The Structured Diffusion Feature Extractor (SDFE) uses anisotropic diffusion to preserve boundary-sensitive features during downsampling. The Mamba-driven Receptive-field Context Aggregator (MRCA) performs multi-directional selective state-space scanning to capture long-range context with linear complexity. The Mask-guided Bayesian Box Refinement (MBBR) applies a MAP-inspired confidence-adaptive box update using MobileSAM mask evidence and ELBO-based false-positive filtering. Using YOLOv13-S as the main detector, SMMNet achieves 32.8% mAP@0.5:0.95 and 52.6% mAP@0.5 on VisDrone2019 at 87 FPS on an NVIDIA A800 GPU, improving the YOLOv13-S baseline by 3.6 and 4.5 points, respectively. The added modules reduce throughput compared with the detector-only baseline (168 FPS), but the resulting 87 FPS remains real-time and provides a favorable accuracy–latency trade-off. Three independent-seed runs further show a mean paired gain of 3.60 ± 0.10 mAP on VisDrone2019, 2.53 ± 0.12 mAP on DroneVehicle, and 2.77 ± 0.06 mAP on SeaDronesSee for the YOLOv13-S setting. Additional experiments on DroneVehicle and SeaDronesSee, together with cross-backbone evaluations on YOLOv5/v6/v7/v8/v10/v11/v13 across different UAV benchmarks, show aligned performance trends under matched settings. Edge deployment on an NVIDIA Jetson Orin NX reaches 30 FPS under TensorRT FP16 inference at 15 W TDP, indicating the suitability of SMMNet for resource-constrained UAV perception.

Abstract Overharvesting of wild medicinal plants threatens biodiversity, rural livelihoods, and the long-term sustainability of harvested populations worldwide. Wild American ginseng (Panax quinquefolius L.) is listed on Appendix II of the Convention on International Trade in Endangered Species (CITES) and, as a result, is regulated in the United States under a federally coordinated, age-based harvest framework intended to sustain natural populations. However, age-based criteria assume strong, consistent relationships between plant age, size, and reproductive capacity—relationships that do not hold up for many other plant and animal species, and have been questioned for American ginseng. We test whether size-based metrics provide a stronger and more practical basis for regulating ginseng harvest than chronological age. We measured age, aboveground traits, root biomass, and reproductive output for wild-simulated ginseng plants (5–10 years old) across four sites spanning substantial environmental heterogeneity. Leaf area, estimated from a simple field measurement, explained 75% of the variation in root dry mass, far exceeding the explanatory power of age (12%) or the combination of age and leaf count (36%), representing the current regulatory approach. Leaf area also strongly predicted the probability of reproduction and maximum seed production, with larger plants consistently outperforming smaller individuals, regardless of age. Our results demonstrate that age-based harvest criteria are weak proxies for the biological attributes most relevant to conservation and sustainability. A simple, field-measurable size metric better reflects ginseng performance and could underpin a transition to size-based regulation, analogous to widely used approaches in fisheries management.