Search Results (3,712)

Arid Central Asia (ACA) is a major source of atmospheric dust in the Northern Hemisphere; however, the evolutionary models and driving mechanisms of Holocene aeolian activity in this region remain debated. Based on 13 reliable AMS ¹⁴C dates from the Sayram Lake SLM2009 sediment core, this study reconstructs the Holocene sequence in aeolian activity through end-member modeling analysis (EMMA). It evaluates its relationship with regional atmospheric circulation. Four end-members were identified from base to top: EM1, with a modal grain size of 7.58 μm, represents low-energy suspension deposition; EM2 (26.30 μm) reflects lacustrine hydrodynamic processes; while EM3 (52.48 μm) and EM4 (416.86 μm) serve as proxies for regional aeolian activity. The results indicate that aeolian activity was relatively strong during the early Holocene (reaching peaks at 11.7–11.2 and 9.2–8.1 cal ka BP), significantly intensified during the mid-Holocene (7.3–5.3 cal ka BP), and gradually weakened in the late Holocene (since 4.0 cal ka BP). Comparison of the aeolian record from Lake Sayram with Greenland ice cores, North Atlantic ice-rafted debris events, and the GISP2 K⁺ record indicates that variations in aeolian activity in arid Central Asia are closely linked to the Northern Hemisphere climate system. We propose that these variations were primarily modulated by large-scale atmospheric circulation, driven by the synergistic interaction between the Siberian High and the mid-latitude westerlies. Full article

Smart-meter theft detection requires learning from fine-grained electricity consumption data, whose centralized processing poses significant privacy risks. Federated learning (FL) mitigates these risks by decentralizing training, but providing rigorous user-level differential privacy (DP) under non-IID data and heterogeneous client behavior remains challenging. Existing DP-FL approaches rely on fixed global clipping bounds for client updates, which substantially overestimate sensitivity when privacy loss is composed using Rényi Differential Privacy (RDP), zero-Concentrated DP (zCDP), or Moments Accountant (MA) frameworks, leading to excessive noise and degraded utility. This work proposes an adaptive clipping-based RDP accountant that incorporates empirical, round-wise update magnitudes into privacy accounting by rescaling each round’s RDP contribution according to the observed clipping ratio. The method is optimizer-agnostic and is evaluated with FedAvg, FedProx, and SCAFFOLD on the SGCC smart-meter theft dataset under IID and Dirichlet non-IID partitions. Experimental results show consistently tighter privacy bounds and improved model utility compared to classical DP accountants, demonstrating the effectiveness of sensitivity-aware privacy accounting for practical differentially private FL. Full article

Climate change is intensifying hydroclimatic extremes and agricultural water scarcity, sharpening trade-offs among yield stability, water saving, and farm incomes in major grain regions. Existing studies often optimise cropping patterns or irrigation schedules separately, seldom embedding yield robustness and policy instruments in one decision framework. We propose an integrated Machine-learning–System-dynamics–Non-dominated-sorting-genetic-algorithm-II (ML–SD–NSGA-II) framework linking long-horizon meteorological scenario generation, crop–water–economy feedback and multi-objective optimisation of crop areas and irrigation depths. ML models generate daily climate sequences to drive an SD model of soil moisture, yield formation, basin-scale allocable water, and farm returns; NSGA-II searches Pareto-optimal strategies that maximise profit and irrigation water productivity while minimising yield deviation. Applied to a rice–wheat irrigation system in the middle Yangtze River Basin, knee-point solutions lift irrigation water productivity by about 14%, maintain near-baseline profits, and reduce yield deviation. Scenario tests with block tariffs, quota-based subsidies, and extreme drought show pricing mainly curbs low-value water use in normal years, while under drought, physical scarcity dominates and economic tools offer limited buffering. This reveals the existence of a scarcity-regime threshold beyond which economic instruments become second-order relative to binding biophysical constraints. The framework supports transparent ex ante testing of tariff–subsidy packages for irrigation governance and adaptation. Full article

The Western Songnen Plain, a critical yet ecologically fragile grain-producing area, is facing sustainability risks arising from rapid land use changes, which demand scientific assessment and regulation. From an ecological security standpoint, this study synthesizes multiple data sources, including GlobeLand30 data, climate, topography, and soil data. Based on the assessment of water conservation, soil conservation and biodiversity maintenance, combined with minimum cumulative resistance model (MCR) and the CLUMondo model, this study comprehensively reveals the dynamic evolutionary patterns of land use in the Western Songnen Plain over the past two decades, concurrently analyzed the spatial heterogeneity pattern of ecosystem services, and further simulated land use changes under natural growth, farmland protection, and ecological security scenarios. According to the results, the grassland area decreased significantly, while cropland and construction land continued to expand. Water conservation, soil conservation, and habitat quality displayed remarkable regional differences, with high values predominantly situated in wetlands, grasslands, and mountainous regions. In contrast, low values exhibited strong spatial correspondence with regions of heightened anthropogenic disturbance. Although the cropland protection scenario promoted agricultural intensification, it reduced ecological heterogeneity. In contrast, the ecological security scenario achieved a higher patch density (0.408) and landscape diversity (1.142) compared to the natural growth scenario, with moderate increases in aggregation. This study identified 27 ecological pinch points, 24 ecological barrier points, and 97 ecological corridors, which provide direct support for regional water and soil resource protection and further underpin the constructed ecological security pattern of “two belts, three zones, and multiple nodes”. These findings have important reference significance for optimizing regional land use structure and maintaining the stability of terrestrial ecosystems in the Western Songnen Plain. Full article

As a crucial pathway for optimizing land factor allocation, farmland transfer plays a pivotal role in implementing the “storing grain in land and technology” strategy and safeguarding national grain security. Based on panel data from 30 provinces in China spanning 2009 to 2023, this study employs a two-way fixed effects model and a Spatial Durbin Model (SDM) to systematically examine the mechanisms, heterogeneity, and spatial spillover effects of farmland transfer on grain production capacity. The results indicate that: (1) Farmland transfer significantly enhances grain production capacity, and this conclusion remains robust after multiple robustness and endogeneity tests. (2) Farmland transfer boosts grain production capacity by promoting cultivated land connectivity and facilitating the substitution of machinery for labor; however, the accompanying non-grain tendency and land governance disputes exert inhibitory effects on capacity release. (3) Transfers to farming households and professional cooperatives, as well as the adoption of leasing and informal exchange arrangements, exhibit the strongest positive effects on production capacity, and the scale-efficiency gains of farmland transfer are particularly pronounced in major grain-consuming areas. (4) Improvements in a region’s farmland transfer level drive the enhancement of grain production capacity in neighboring regions through the diffusion of management experience and the sharing of social services. This study provides empirical evidence and policy insights to optimize farmland transfer mechanisms and safeguard food security. Full article

Chickpea (Cicer arietinum L.) is a globally important legume crop whose grain yield is strongly influenced by environmental and agronomic variability. This study aimed to predict chickpea grain yield using artificial neural networks (ANNs) and to identify key traits associated with yield formation across different genotypes under semi-arid conditions. The dataset consisted of 96 chickpea genotypes evaluated over two growing seasons (2022–2023) in Sivas, Türkiye. The results demonstrated that reproductive traits, particularly seed weight per plant, number of pods per plant, and number of seeds per plant, were the most influential factors determining grain yield. Environmental variability also contributed significantly to yield prediction, highlighting the importance of genotype–environment interactions. The developed ANN model showed high predictive accuracy, indicating its robustness in capturing complex relationships among yield-related traits. Beyond prediction, the model provides biologically meaningful insights into trait prioritization, supporting its application in chickpea breeding programs. Overall, the findings suggest that ANN-based approaches can serve as effective decision-support tools in precision agriculture by enabling accurate yield estimation, facilitating the selection of high-performing genotypes, and identifying key breeding traits for sustainable crop improvement. Full article

Millimeter-wave radar has gradually gained popularity as a sensor mode for Human Activity Recognition (HAR) in recent years because it preserves the privacy of individuals and is resistant to environmental conditions. Nevertheless, the fast inference of high-dimensional and sparse 4D radar data is still difficult to perform on low-resource edge devices. Current models, including 3D Convolutional Neural Networks and Transformer-based models, are frequently plagued by extensive parameter overhead or quadratic computational complexity, which restricts their applicability to edge applications. The present paper attempts to resolve these issues by introducing RadarSSM as a lightweight spatiotemporal hybrid network in the context of radar-based HAR. The explicit separation of spatial feature extraction and temporal dependency modeling helps RadarSSM decrease the overall complexity of computation significantly. Specifically, a spatial encoder based on depthwise separable 3D convolutions is designed to efficiently capture fine-grained geometric and motion features from voxelized radar data. For temporal modeling, a bidirectional State Space Model is introduced to capture long-range temporal dependencies with linear time complexity $O\left(T\right)$, thereby avoiding the quadratic cost associated with self-attention mechanisms. Extensive experiments conducted on public radar HAR datasets demonstrate that RadarSSM achieves accuracy competitive with state-of-the-art methods while substantially reducing parameter count and computational cost relative to representative convolutional baselines. These results validate the effectiveness of RadarSSM and highlight its suitability for efficient radar sensing on edge hardware. Full article

Hyperspectral image (HSI) classification regarding vegetation is hampered by strong intra-class spectral variability and inter-class similarity, and commonly used random pixel splits can introduce spatial-context leakage that inflates test accuracy in patch-based models. To address these issues, we propose a classification framework that couples a leakage-free block partition (LFBP) strategy with class-aware multi-prototype contrastive loss (CAMP-CL). LFBP assigns non-overlapping spatial blocks to training/validation/test sets and reserves a buffer matched to the patch radius to prevent contextual overlap while keeping class distributions balanced. CAMP-CL represents each class with multiple learnable prototypes and performs supervised contrastive learning at the prototype level, encouraging compact yet multimodal intra-class embedding and improved inter-class separation. Experiments conducted on the Matiwan Village airborne HSI dataset under the LFBP protocol show that the proposed method can achieve 91.51% overall accuracy (OA) and 91.49% average accuracy (AA). Compared with the strongest baseline, supervised contrastive learning (SupCon), the proposed method yields consistent gains of 1.07 percentage points (pp) in both OA and AA while improving OA by 5.76 pp over the cross-entropy baseline. The results suggest that CAMP-CL is beneficial for addressing the challenges of HSI classification for fine-grained vegetation, while leakage-free evaluation protocols are important for obtaining more reliable performance estimates in practical settings. Full article

Integrating agricultural remote sensing and phenomics for full-growth-period rice quality prediction is vital for early non-destructive screening and breeding; however, studies integrating genomic and multi-source phenotypic data across multiple environments remain limited. This study addressed this gap by integrating genomic SNP data, UAV-based spectral data, and individual multidimensional phenotypic data of 61 indica rice varieties (field and greenhouse environments). As a proof-of-concept study, feature selection methods (LASSO, MI, RFE, SPA) were used to mitigate overfitting and the “p >> n” problem, with further validation needed in larger populations. The results showed that amylose content is genetically dominated, protein content is genetically determined and influenced by gene-environment interactions, and chalkiness traits are determined by three combined factors. For amylose content, SNP data under the Random Forest model at the population level (phenomics data from field UAV remote sensing of variety populations) achieved optimal performance (R² = 0.92; MAE = 1.1; RMSE = 1.5), while the Stacking Ensemble method enhanced accuracy at the individual level (phenomics data from greenhouse single-plant phenotyping per variety). Chalky grain rate and chalkiness degree showed SNP-comparable prediction accuracy, with Stacking significantly improving performance at the population level (R² = 0.89 and 0.85, respectively). Protein content prediction remained relatively low (optimal R² = 0.56) due to strong environmental sensitivity and complex interactions. This framework extends traditional single-environment/single-data-source approaches, providing an effective strategy for early, high-throughput, non-destructive rice quality screening. Further validation with larger datasets, more growing seasons, or independent populations is required for reliable application in breeding-related practices. Full article

As water resources are becoming increasingly scarce in the North China Plain, irrigation strategies that simultaneously improve grain yield and reduce irrigation water input are needed for winter wheat (Triticum aestivum L.) production. Current irrigation decision rules are based either on fixed soil moisture thresholds or on evapotranspiration (ET)-based ratios applied uniformly across the growing season, limiting their flexibility for growth stage-specific irrigation management. In this study, a multi-objective simulation optimization framework was developed to jointly optimize soil moisture lower control limits (irrigation trigger thresholds) and evapotranspiration-based irrigation replenishment ratios across key winter wheat growth stages. The framework integrated the AquaCrop-OSPy crop model with the PyFAO56 soil moisture balance, irrigation scheduling model and the NSGA-II evolutionary optimization algorithm. A field experiment was conducted during the 2024–2025 growing season in Laoling City, Shandong Province, China, employing a four-dense–one-sparse strip cropping pattern with two irrigation treatments: T1 (subsurface sprinkler irrigation) and T2 (shallow subsurface drip irrigation). The AquaCrop-OSPy model was calibrated and validated using measured canopy cover, aboveground biomass, grain yield, and soil moisture content in the 0–60 cm soil layer. Simulated canopy cover and grain yield showed good agreement with observations, with the coefficient of determination (R²) ranging from 0.87 to 0.94. For grain yield, the normalized root mean square error (NRMSE) ranged from 2.24% to 3.75%, and the root mean square error (RMSE) ranged from 0.29 to 0.54 t·ha⁻¹. For aboveground biomass, R² was 0.99, while RMSE ranged from 1.02 to 1.11 t·ha⁻¹, and NRMSE ranged from 14.25% to 15.49%. The PyFAO56 irrigation strategy model simulated average root-zone soil-moisture dynamics with satisfactory accuracy, with an R² of 0.86 and an RMSE of 5%. Multi-objective optimization (maximizing yield while minimizing irrigation volume) generated 23 Pareto-optimal irrigation strategies, with irrigation volumes ranging from 51 to 128 mm, corresponding yields ranging from 9.8 to 10.8 t·ha⁻¹, and irrigation water use efficiency (IWUE) ranging from 0.08 to 0.19 t·ha⁻¹·mm⁻¹. Correlation analysis within the Pareto set indicated that soil-moisture control lower limits during the regreening–jointing stage and higher soil-moisture control lower limits during the flowering–maturity stage were key controlling factors for achieving high yields and irrigation water use efficiency. The Entropy-Weighted Ranked Minimum Distance method identified an optimal irrigation scheme involving two irrigations (one at the end of the jointing stage and another at the beginning of the grain filling stage) involving an irrigation depth of 75 mm, achieving a simulated yield of 10.4 t·ha⁻¹ and an IWUE of 0.16 t·ha⁻¹·mm⁻¹. The proposed AquaCrop-PyFAO56-NSGA-II framework provides a flexible, process-based workflow for jointly optimizing irrigation control thresholds and evapotranspiration-based irrigation replenishment ratios across different winter wheat growth stages. Under the monitored conditions of the 2024–2025 wet season, the framework identified a two-irrigation strategy that balanced grain yield and irrigation input. This study should, therefore, be regarded as a proof-of-concept evaluation conducted in a well-instrumented single-site field setting rather than as a universally transferable recommendation. Because model calibration, within-season validation, and optimization were all based on one wet growing season at one site, the derived stage-specific thresholds, Pareto front, and S5 recommendation are most applicable to hydro-climatic conditions similar to the study year and should be further tested across contrasting year-types and locations before broader extrapolation. Full article

Quantitative microstructure evaluation of sedimentary rock thin sections is essential for revealing reservoir flow mechanisms and assessing reservoir quality. However, traditional manual identification is inefficient and prone to subjectivity. Although current deep learning approaches have improved efficiency, most remain confined to single tasks and lack a pathway to translate image recognition into quantifiable geological parameters. Moreover, these methods struggle with cross-scale feature extraction and accurate grain boundary localization in complex textures. To overcome these limitations, this study proposes a three-stage automated analysis framework integrating intelligent lithology identification, sandstone grain segmentation, and quantitative analysis of fabric parameters. To address scale discrepancies in lithology discrimination, Rock-PLionNet integrates a Partial-to-Whole Context Fusion (PWC-Fusion) module and the Lion optimizer, which mitigates cross-scale feature inconsistencies and enables accurate screening of target sandstone samples. Subsequently, to correct boundary deviations caused by low contrast and grain adhesion, the PetroSAM-CRF strategy integrates polarization-aware enhancement with dense conditional random field (DenseCRF)-based probabilistic refinement to extract precise grain contours. Based on these outputs, the framework automatically calculates key fabric parameters, including grain size and roundness. Experiments on 3290 original multi-source thin-section images show that Rock-PLionNet achieves a classification accuracy of 96.57% on the test set. Furthermore, PetroSAM-CRF reduces segmentation bias observed in general-purpose models under complex texture conditions, enabling accurate parameter estimation with a roundness error of 2.83%. Overall, this study presents an intelligent workflow linking microscopic image recognition with quantitative analysis of geological fabric parameters, providing a practical pathway for digital petrographic evaluation in hydrocarbon exploration. Full article

Wire and arc additive manufacturing (WAAM) offers high deposition efficiency for large-scale aluminum components; however, layer-by-layer thermal cycling often induces microstructural anisotropy and spatial heterogeneity, which compromise structural reliability. In this study, an ultrasonic-based quantitative framework is proposed to evaluate and optimize anisotropy and heterogeneity in WAAM 2319 aluminum alloy. Nine blocks were fabricated using an orthogonal design with three key process parameters: torch travel speed, arc current, and shielding gas flow rate. Ultrasonic velocity and attenuation were employed to construct anisotropy and heterogeneity indicators. Results show that velocity-based anisotropy remains below 0.53%, indicating nearly isotropic elastic stiffness, whereas attenuation-based anisotropy reaches up to 76%, revealing pronounced direction-dependent microstructural and porosity features. Metallographic analysis confirms that grain morphology variation and interlayer porosity jointly govern attenuation responses. Response surface surrogate models were established to correlate ultrasonic indicators with process parameters, and both single- and multi-objective optimizations were performed within the feasible process window. The proposed framework provides a non-destructive, volumetric approach for microstructure-informed process parameter optimization in WAAM aluminum alloys. Full article

This paper investigates the control of magnetic energy confinement in one- and three-dimensional magnetic systems by systematically accounting for magnetic interactions. The analysis provides new insight into magnetic behavior at the nanoscale and introduces a simulation-based framework that clearly distinguishes between magnetizing and demagnetizing interaction regimes. Within this framework, magnetic energy confinement is rigorously defined and can be quantitatively controlled through the underlying interaction landscape. To validate this approach, extensive numerical simulations were performed on representative one- and three-dimensional nanostructures, including individual nanowires and hexagonal arrays of nanowires. Each nanowire was modeled as a chain of interacting ellipsoidal grains, enabling an accurate description of the complex magnetic interactions governing energy confinement in these systems. Full article

The selection of deformation mechanisms in hexagonal close-packed (HCP) metals is strongly influenced by both crystallographic orientation and macroscopic deformation constraints. In commercially pure titanium, plastic deformation under constrained loading conditions involves a complex interplay between dislocation slip and deformation twinning, whose respective activation cannot be fully described by classical stress-based criteria. In this study, the mechanical origin of slip and twinning variant selection in commercially pure titanium subjected to plane strain compression is investigated experimentally. Plane strain compression is used as a canonical loading condition representative of constrained deformation paths encountered in sheet metal forming. Interrupted in-situ electron backscatter diffraction is combined with slip trace and twin variant analyses to identify the active deformation mechanisms at the grain scale. Resolved shear stress calculations show that stress-based criteria provide a necessary first-order condition for the activation of both slip and twinning systems. While the Schmid factor reasonably predicts part of the observed slip activity, it fails to uniquely determine the selection of active twinning variants. A kinematic analysis reveals that twinning variant selection is governed by the compatibility between the deformation induced by twinning and the macroscopic strain constraints imposed by plane strain compression. Only variants whose deformation accommodates compression along the loading axis, extension along the free in-plane direction, and minimal strain along the constrained in-plane direction are preferentially activated. These results demonstrate that deformation mechanism selection in HCP titanium under constrained loading conditions results from a combined effect of resolved shear stress and kinematic compatibility. The proposed framework provides a physically grounded basis for interpreting deformation-induced texture evolution and offers clear perspectives for the development of crystal plasticity models incorporating twinning under complex strain paths. Full article

Hyperspectral image (HSI) classification is a core task in remote sensing data interpretation. Although recently introduced state space models (SSMs), such as Mamba, have demonstrated promising performance in hyperspectral analysis due to their linear computational complexity and strong long-sequence modeling capability, existing single-stream scanning mechanisms struggle to effectively balance the intrinsic spectral continuity dependency and the high-dimensional redundancy inherent in HSI data. Moreover, they often suffer from representation discrepancies when fusing features from heterogeneous representation spaces. To address these challenges, we propose a continuous–discrete collaborative framework, termed Confluence Mamba (CF-Mamba). Specifically, the continuous modeling path (AHSE) introduces a multi-view adaptive routing mechanism to accurately capture anisotropic spectral–spatial continuous evolution patterns. Simultaneously, the discrete interaction path (IISE) employs interval sampling and channel shuffling strategies to efficiently decouple high-dimensional redundancy while maintaining fine-grained feature interactions. Furthermore, the confluence gating unit (CGU) leverages a bidirectional cross-modulation mechanism to constrain discrete feature distributions using continuous contextual information, effectively alleviating representation discrepancies during multi-scale feature fusion. Extensive experiments conducted on four benchmark datasets, namely, Indian Pines, Pavia University, Houston, and WHU-Hi-Longkou, demonstrate that CF-Mamba achieves overall accuracies of 97.77%, 99.68%, 99.06%, and 99.59%, respectively. The proposed method consistently outperforms existing CNN-, Transformer-, and Mamba-based approaches in terms of both classification performance and computational efficiency. Full article