TerraMosaic Daily Digest: September 9, 2026
Daily Summary
Two studies move InSAR landslide analysis beyond deformation detection. A graph-based framework first extracts candidate deformation patches, then combines spatial connectivity with ten terrain, geological, hydrological and anthropogenic factors to distinguish active landslides from engineering, agricultural and subsidence signals in the Bailong River Basin. A complementary analysis of 498 active landslides on the eastern Tibetan Plateau shows that longitudinal line-of-sight velocity profiles encode type-dependent kinematics and internal deformation zones, enabling first-order inference of movement style and failure-surface geometry where subsurface observations are sparse. Region-similarity assessment addresses the transfer problem upstream: a confidence-ellipse index identifies source domains most compatible with a data-poor target before physics-informed susceptibility models are trained.
Material contrasts and hydrological history dominate the day's process studies. Ring-shear experiments on ice-rock analogues show that increasing the high-friction fraction raises peak and residual strength, dilatancy and apparent viscosity, whereas ice-rich mixtures contract and promote basal lubrication. In the Three Gorges region, continuum-discontinuum simulations identify toe-confined, progressive and complete failures of deteriorating anti-dip slopes; the transition depends strongly on the combined bedding and slope angles and on repeated wetting-drying within the reservoir fluctuation belt. Swedish river studies show why climate attribution is not straightforward: discharge resolution materially changes projected erosion, and hydropower regulation can produce stronger erosion than climate-driven flow change. Along the Jinsha River, GNSS and inclinometer observations place an active sliding surface at 22-34 m depth and identify rapid drawdown as a critical destabilizing condition.
Risk models are increasingly organized around function rather than physical damage alone. A Bayesian network maps incomplete observations of earthquake and coseismic-landslide damage into graded transportation functionality, updating conditional probabilities with post-event evidence from Wenchuan mountain tunnels. El Salvador's vertical-motion hazard maps extend probabilistic seismic assessment across multiple return periods, while a Nature Geoscience synthesis across 15 subduction zones finds no relation between interface roughness and maximum earthquake magnitude. In mountain river networks, D-CASCADE simulations indicate that less than 1% of a debris-flow sediment pulse typically reached the basin outlet over nearly a decade, with long-lived storage concentrated below source zones and slope breaks. Across wildfire, flood and drought studies, the common advance is explicit treatment of spatial heterogeneity, compound forcing and the operational quantity that a decision actually requires.
Key Trends
The day's methods convert remotely observed change into process interpretation, transferable models and functional risk estimates.
- InSAR is moving from anomaly mapping to geomechanical inference: Graph inference rejects non-landslide deformation by combining InSAR with environmental priors, while longitudinal velocity profiles distinguish movement regimes and internal zones across hundreds of active slopes. The target is now mechanism and geometry, not detection alone.
- Transfer learning is becoming region-aware: Confidence-ellipse similarity provides an explicit pre-training test for choosing source regions in landslide susceptibility mapping. Related remote-sensing methods emphasize domain alignment, multimodal registration and uncertainty rather than assuming that large source datasets transfer uniformly.
- Material heterogeneity governs long-runout and progressive failure: Ice-rock friction contrast regulates avalanche strength, dilation and lubrication; reservoir-band deterioration reorganizes failure in anti-dip slopes; and water-level cycling controls deep-seated deformation along the Jinsha River.
- Hazard models are coupling forcing histories to network consequences: River erosion projections retain regulation and discharge resolution, debris-flow sediment is tracked through decadal network storage, and compound flood-waterlogging studies connect climate forcing to urban exposure and infrastructure performance.
- Functional fragility is replacing damage-only assessment: Bayesian updating translates component damage under shaking and coseismic landslides into transportation functionality. Flood-resilience, warning and infrastructure studies similarly optimize the service or action that must persist after the event.
Selected Papers
The September 9 collection is anchored by active-landslide graph inference, LOS-profile reconstruction of slope kinematics and failure geometry, region-aware transfer learning, ice-rock avalanche friction and reservoir-slope deterioration. Companion studies address climate-sensitive river erosion, debris-flow sediment routing, earthquake and coseismic-landslide infrastructure fragility, seismic hazard, GNSS slope monitoring, wildfire mapping and compound urban flooding. The methodological layer spans multimodal Earth observation, physics-informed learning, uncertainty-aware forecasting, geotechnical sensing and network resilience, linking observed change to mechanism and operational consequence.
1. A Graph-Based Inference Framework for Active Landslide Identification by Integrating InSAR Data and Landslide Prior Knowledge
Core Problem: InSAR deformation candidates include engineering, agricultural and subsidence signals that can be mistaken for active slopes.
Key Innovation: Combines YOLO candidate extraction with a spatial-environmental graph and ten prior-factor groups, allowing a GNN-transformer to distinguish landslide-related deformation in the Bailong River Basin.
2. Maximum earthquake magnitude unlikely to be controlled by subduction interface roughness
Core Problem: Subduction-interface roughness is widely proposed as a control on maximum earthquake size, but its global empirical support is uncertain.
Key Innovation: Compares roughness metrics and maximum magnitude across 15 subduction zones and finds no relationship, challenging topography-based limits on megathrust potential.
3. Inferring landslide kinematics and failure surface geometry from longitudinal LOS velocity profiles
Core Problem: Remotely sensed deformation rarely constrains subsurface failure geometry in mountains with sparse field observations.
Key Innovation: Uses diagnostic longitudinal LOS profiles, conceptual models and finite-element references across 498 active landslides to recover type-dependent motion regimes and internal deformation zones.
4. Probabilistic Functional Fragility Assessment of Transportation Infrastructure Under Seismic and Coseismic Landslide Hazards: A Data-Driven Bayesian Network Approach
Core Problem: Physical fragility does not directly quantify how incomplete, heterogeneous component damage translates into system functionality.
Key Innovation: Maps PGA and localized Newmark displacement through an evidence-updated Bayesian network to graded functional loss, validated with Wenchuan mountain-tunnel observations.
5. The role of friction difference between ice and rock on mobility of ice-rock avalanches based on ring shear tests
Core Problem: Runout models do not quantitatively resolve how low-friction ice and high-friction rock interact over large shear displacement.
Key Innovation: Ring-shear tests isolate friction heterogeneity, showing nonlinear increases in strength, dilation and viscosity with rough-particle content and contraction-driven lubrication in ice-rich mixtures.
6. Region similarity assessment for empowering physics-informed transfer learning-based landslide susceptibility mapping
Core Problem: Transfer learning can degrade when the source landslide region is poorly matched to a data-scarce target.
Key Innovation: Introduces a confidence-ellipse similarity index for source-region selection and shows susceptibility-transfer performance increasing with measured regional similarity.
7. Projecting climate-induced river erosion for assessing future landslide susceptibility: methodological insights and lessons learned from five Swedish cases
Core Problem: Future slope-risk assessments often omit how regulation, discharge resolution and channel change alter erosion forcing.
Key Innovation: Synthesizes five Swedish applications into an adaptive cross-disciplinary workflow, showing that hydropower operations can obscure or exceed climate-driven erosion signals.
8. GNSS monitoring and stability evolution mechanism of the S-J1 deformation body in the lower reaches of the Jinsha River
Core Problem: Large reservoir banks can remain marginally stable while progressive deformation develops below the surface.
Key Innovation: Combines GNSS, inclinometers and stability analysis to locate a 22-34 m sliding surface and identify reservoir drawdown, seismic loading and seepage as critical controls.
9. Debris flow-derived sediment routing and storage in a mountain river network: a D-CASCADE application in the Trebbia-Aveto River (Italy)
Core Problem: The residence time and downstream export of debris-flow sediment pulses remain difficult to quantify at basin scale.
Key Innovation: D-CASCADE tracks nearly a decade of routing under five transport equations, finding dominant in-basin storage and typically less than 1% export at the basin outlet.
10. Monitoring of fractured rock bolted slopes using intelligent terminal structure
Core Problem: Conventional bolted-slope monitoring does not directly capture shear deformation at anchored joints under extreme loading.
Key Innovation: Uses tooth-fracture responses in an intelligent terminal structure, laboratory shear tests and field IoT measurements to monitor anchored rock-joint deformation during cyclonic rainfall.
11. Failure behaviour of soft-hard interbedded anti-dip rock slopes due to rock deterioration using continuum-discontinuum element method
Core Problem: Repeated wetting and drying in reservoir fluctuation belts can reorganize failure across soft-hard interbedded anti-dip slopes.
Key Innovation: Identifies three failure scenarios and geometry-dependent triggering versus inducing mechanisms, linking complete failure to rock deterioration and combined bedding-slope angle.
12. The Spatiotemporal Evolution of Creep on the Southern San Andreas Fault Between 2015-2021
Core Problem: Although average creep rates are relatively well-constrained, untangling the details of spatiotemporal variation in creep along the SSAF has been challenging due to limited data coverage.
Key Innovation: In this study, we use 7 years of Interferometric Synthetic Aperture Radar data from the dense Sentinel-1 catalog to image and model the evolution of shallow creep along the SSAF.
13. Infra-Bench CLS: A Global, Open-Source Benchmark for Critical Infrastructure Classification with Earth Observation Foundation Models
Core Problem: Yet, foundation models remain largely untested for detecting and classifying the facility-scale critical infrastructure that underpins a range of important societal and economic functions.
Key Innovation: Subsequently, Infra-Bench CLS is introduced as a benchmark to test foundation models on 18,756 Sentinel-1 SAR and Sentinel-2 multispectral facility-scale critical infrastructure asset images covering seven continents and 13 infrastructure classes, with results reported for the 10 retained classes.
14. Multi-Agent Reinforcement Learning for Autonomous UAV Exploration in Wildfire Response
Core Problem: This study develops a deep reinforcement learning framework for training Unmanned Aerial Vehicle (UAV) agents to navigate and monitor simulated wildfire environments.
Key Innovation: This study develops a deep reinforcement learning framework for training Unmanned Aerial Vehicle (UAV) agents to navigate and monitor simulated wildfire environments. Results show that agents learn increasingly stable and effective behaviors over time, as demonstrated by converging loss trends, improved reward signals, and more consistent navigation patterns such as fire-boundary tracking.
15. Collocated Geotechnical and Geophysical Test Dataset at Three Sites in Napier, New Zealand
Core Problem: Reliable assessment of liquefaction hazard, particularly in the context of groundwater fluctuations associated with hydroclimatic changes and variability between liquefaction assessment methods, necessitates supplementary data and advanced site characterization methods.
Key Innovation: To understand liquefaction hazard and its relation to nonstationary groundwater levels in Napier, Hawke's Bay, New Zealand, a two-part field testing campaign was conducted at three sites during two distinct seasons, dry (November 2023) and wet (June 2024).
16. Integrating Nature-Based Solutions (NbS) for Enhanced Flood Resilience under a Changing Climate: The Case of the Cologne District, Germany
Core Problem: Abstract.
Key Innovation: This study addresses this gap by mapping, categorising and evaluating existing and planned NbS for flood risk mitigation. The results demonstrate that multiple NbS have been implemented and are planned along the Rhine, but additional efforts are needed in the Erft and Wupper tributaries, despite several planned and implemented river restoration projects.
17. Development of a high-resolution coupled SHiELD-MOM6-LM4 model - Part 2: Model overview, coupling technique, and evaluation of hydrological extremes during Hurricane Helene
Core Problem: This work describes the implementation strategy and technical challenges involved in integrating the Geophysical Fluid Dynamics Laboratory (GFDL)’s Land Model (LM4) with dynamic subgrid tiling capabilities within the atmospheric model, System for High-resolution modeling for Earth-to-Local Domains (SHiELD), capable of kilometer-scale global simulations.
Key Innovation: A key challenge addressed in this effort is coupling LM4, which was designed for implicit surface flux coupling, with SHiELD’s explicit physics solver. We achieve this through a refactoring of the atmospheric physics suite and code drivers, enabling implicit land-atmosphere coupling of heat and moisture within the well-established FMS coupler infrastructure.
18. Multilayer soil moisture deficit amplifies drought impacts on global ecosystems
Core Problem: Nature Geoscience, Published online: 10 September 2026; doi:10.1038/s41561-026-02082-2 Our global analysis of multi-source, multilayer soil moisture datasets shows that droughts become particularly damaging when moisture deficits occur simultaneously throughout the soil profile, eliminating vertical hydrological buffering.
Key Innovation: These vertically compound droughts have intensified across much of the globe and pose a growing threat to carbon uptake in forests and croplands.
19. Wildfire Scar Detection in Mediterranean Chile Using Sentinel-1 InSAR Coherence and Machine Learning: The 2017 “Las Máquinas” Megafire
Core Problem: Wildfire monitoring using synthetic aperture radar (SAR) provides critical capabilities under challenging atmospheric conditions where optical sensors are limited by smoke and cloud cover.
Key Innovation: We evaluated Sentinel-1 C-band SAR interferometric coherence for Burned-area detection of the 2017 “Las Máquinas” megafire (Maule, Chile), comparing Ascending (Asc) and Descending (Dsc) orbital geometries processed with the AMSTer InSAR software. For the Ascending orbit, XGBoost achieved the highest performance (OA = 0.9328; F1 = 0.9195) and mapped 143,950 ha (76.2%) as Burned.
20. Coastal risk assessment and hazard forecast analysis via a Bayesian network
Core Problem: In this study, a Bayesian network (BN) module is proposed for coastal hazard risk assessment and forecasting based on the statistical prior and conditional probabilities, which capture the mechanisms of transitivity and causality.
Key Innovation: In this study, a Bayesian network (BN) module is proposed for coastal hazard risk assessment and forecasting based on the statistical prior and conditional probabilities, which capture the mechanisms of transitivity and causality. The predictive accuracy of the module is approximately 74%, demonstrating its substantial practical value for coastal planning.
21. Seismic hazard maps for El Salvador: the vertical component of motion
Core Problem: This article presents time-independent probabilistic seismic hazard maps for El Salvador in terms of the vertical component of motion regarding the peak ground acceleration and spectral ordinates for 0.2 and 1 s of 5% of critical damping setting 50, 95, 475, 975, and 2475 years return period at rock site conditions and flat topography.
Key Innovation: The area source and smoothed seismicity methods are employed in the assessment.
22. Extraction of knickpoint series based on high-resolution DEM and identification of paleo-earthquakes: A case study of the Yuguang Basin Boundary Fault in the Shanxi Rift, China
Core Problem: Knickpoints tend to form in the upstream reaches of rivers crossing active fault zones.
Key Innovation: This study aims to validate the applicability of knickpoint analysis for paleoseismic reconstruction along the Yuguang Basin Boundary Fault (YBBF), a major active boundary fault in the northern Shanxi Rift, China. These results indicate that the YBBF poses considerable seismic hazard risks that merit attention in future assessments.
23. Unveiling the effects of multidimensional urban shrinkage on urban flood resilience: A case study of the Yangtze River Basin, China
Core Problem: Climate change poses growing flood-management challenges for cities undergoing economic, demographic, and spatial decline.
Key Innovation: Using 1921 city-year observations for 113 prefecture-level cities in China’s Yangtze River Basin (YRB) from 2007 to 2023, this study develops a six-subsystem urban flood resilience (UFR) framework encompassing socioeconomic conditions, ecological environment, infrastructure, institutional support, spatial structure, and climate stress.
24. Interpretable rockburst intensity prediction using a criterion-similarity decoupled fuzzy-regularized graph convolutional network
Core Problem: Accurate prediction of rockburst intensity levels is critical for the safety of deep underground engineering construction.
Key Innovation: To overcome the information loss and instability caused by hard threshold division, a fuzzy regularization mechanism was introduced, mapping discrete criterion features to continuous membership degree vectors and enhancing prediction consistency through a smoothing regularization term. The five-fold cross-validation yields an average accuracy of 94.42%, demonstrating the superior predictive performance of the FR-GCN model.
25. Projected intensification of compound river flooding and urban waterlogging in riverside cities of the Yangtze river basin
Core Problem: However, quantitative assessment of these compound flooding and waterlogging events (CFWE) remains challenging due to their complex dynamics and spatial heterogeneity.
Key Innovation: To address this, we propose a novel two-dimensional Composite Flooding and Waterlogging Index (CFWI) based on a nonstationary Copula-based framework incorporating antecedent hydrological conditions, which integrates both the severity and pattern of CFWE enabling identification of dominant contributions and facilitating consistent cross-regional comparison. Results show that both the frequency and severity of CFWE increase.
26. Hillslope lag times for flood runoff are extremely short in high-relief meso-scale catchments
Core Problem: However, hydrological observations during heavy rainfall events in high-relief meso-scale catchments remain scarce.
Key Innovation: This study aimed to clarify flood propagation in high-relief meso-scale catchments with long and steep hillslopes. Events showing a significant positive correlation between modal channel length and peak lag time accounted for only up to 15% of all events; in these events, hillslope lag time was estimated to be relatively uniform and extremely short (mean: −0.84 min; maximum: 160 min).
27. Mapping Global GNSS Vertical Velocities to Solid Earth Figure Change
Core Problem: The Earth's figure evolves in response to glacial isostatic adjustment (GIA), surface-mass redistribution, and rotational and mantle-driven processes.
Key Innovation: We develop a geometric framework to estimate Earth figure change (EFC) parameters from global vertical velocity fields and their gridded representations, focusing on the degree-2 zonal component that depicts the rate-of-change of solid Earth's flattening. Results indicate an increase in polar uplift from ∼0.5 to ∼1.0 mm/yr and concurrent equatorial subsidence of similar magnitude, producing an accelerating signal of solid.
28. Beyond SMAP: Can GNSS-R Platforms Ensure Continuity of Global Soil Moisture Observations?
Core Problem: The absence of a definitive follow-on strategy for the Soil Moisture Active Passive (SMAP) mission, which has already surpassed its planned mission lifetime, presents a significant challenge for the hydrological science community.
Key Innovation: In this study, we compared soil moisture products from these two systems across seven perspectives: (a) observation characteristics, (b) spatial resolution, (c) coverage and frequency of observations, (d) vulnerability to radio frequency interference, (e) complementarity potential, (f) performance, and (g) cost considerations.
29. Accountable and uncertainty-aware evaluation of sensor-based AI under distribution shift: devices, subjects, and nearly three years underground
Core Problem: Sensor-based AI systems are rarely operated under the conditions under which they were trained: devices, personnel and recording epochs change, and each change degrades performance in ways a random train-test split cannot reveal.
Key Innovation: We propose a staged, accountable evaluation protocol that treats the evaluation of a deployed model as a measurement with declared reference levels and a quantified uncertainty. We demonstrate the protocol on infrastructure-free geomagnetic localisation with smartphone-based recurrent classifiers in two real underground mines, including a replication of the scheme's training stages at the second site.
30. VANTAGE-Bench: Evaluating the Infrastructure AI Gap in Vision-Language Models
Core Problem: As Vision-Language Models (VLMs) advance toward physical deployment, the focus has remained on action-oriented Embodied AI evaluated on subject-centric consumer video.
Key Innovation: We introduce VANTAGE-Bench, a benchmark measuring this "Infrastructure AI Gap." It spans three operational domains (Logistics, Transportation, and Smart Spaces), unifies image and video evaluation across semantic, spatial, temporal, and spatio-temporal capabilities, and moves beyond multiple-choice to eight task formulations including dense captioning and spatio-temporal grounding. Evaluating 17 models zero-shot, we find the.
31. Uncertainty-Aware Sea-Ice Type Mapping with Multiple Ice Charts
Core Problem: These annotations are not exact, however; this is because chart interpretation relies on analyst judgement and on the observations available at the time, so different ice services may assign different SoD labels to the same conditions.
Key Innovation: SoD labels are obtained from operational ice charts, where trained analysts interpret satellite observations and assign standardized stage codes to regions with similar ice conditions. We observe that supervision incorporating information from multiple annotators can improve this correspondence, with soft supervision achieving the highest overall correlation of 0.256.
32. Hyperbolic Geometry for Open-World Object Detection in Remote Sensing Imagery
Core Problem: To improve unknown object recall, we design a two-step unknown-object discovery mechanism: a Decoupled Objectness Learning (DOL) module that disentangles foreground perception from semantic information to separate foreground proposals from background regions, followed by a Hyperbolic Uncertainty Learning (HUL) component that leverages the radius of hyperbolic embeddings as an uncertainty-aware cue for known-unknown.
Key Innovation: For incremental learning, we develop a Hyperbolic Metric Learning (HML) strategy that enhances inter-class separability, facilitating the incorporation of novel categories while mitigating catastrophic forgetting. Experiments on three remote sensing benchmarks demonstrate consistent improvements in unknown recall and incremental learning over state-of-the-art OWOD methods.
33. MethaneFuse: Learning from Multi-Sensor Satellite Observations for Methane Plume Detection
Core Problem: Methane plume detection from satellite imagery is constrained by incomplete observations: public satellites provide complementary spatial, spectral, and atmospheric evidence, but real plume cases rarely contain fully paired multi-sensor measurements because of revisit schedules, cloud coverage, acquisition quality, and the transient nature of emissions.
Key Innovation: Most learning-based detectors rely on single-sensor inputs, especially Sentinel-2 (S2), leaving many reported plume cases unusable. At the representative 480 m setting, MethaneFuse achieves 84.87 F1 and 93.62 AUROC, improving over the strongest baseline by 5.65 F1 and 8.30 AUROC points while reducing false positives by 8.19 points.
34. From Pixels to Hierarchical Sequences: Quadtree Mask Encoding for Vision-Language Binary Change Detection
Core Problem: Dense change detection in remote sensing requires vision-language models (VLMs) to compare bi-temporal images and generate accurate pixel-level masks.
Key Innovation: We introduce QUAKE-CD, a framework that recasts dense change prediction as syntax-verifiable structured generation. On QUAKE-CoT, QUAKE-CD achieves 78.31% accumulated F1, outperforming decoder-based and flat text-as-mask VLMs while producing more faithful bi-temporal reasoning.
35. CLFTv2: Efficient Camera-LiDAR Fusion for Semantic Segmentation via Hierarchical Feature Pyramids
Core Problem: Semantic segmentation for autonomous driving requires reliable detection of vulnerable road users (VRUs) despite heavy class imbalance.
Key Innovation: We introduce CLFTv2, a hierarchical camera-LiDAR fusion framework replacing global ViT attention with a Swin-based multi-scale encoder and a lightweight FPN-style residual decoder. On ZOD, CLFTv2-Large achieves 53.5% mIoU, improving pedestrian IoU from 35.5% to 44.9% over the prior CLFT model.
36. ScopeMamba-YOLO: Widening the Perceptual Scope Inward and Outward for Small Object Detection in Remote Sensing Imagery
Core Problem: Small object detection in unmanned aerial vehicle (UAV) and remote sensing imagery requires preserving high-resolution detail while modeling long-range context.
Key Innovation: We propose ScopeMamba-YOLO, built around an off-path, zero-gated selective-scanning principle that decouples contextual modeling from the convolutional stream. Controlled experiments show that matched main-path selective scanning reduces mAP50 by 0.98 pp, whereas off-path CGCM improves the final configuration by 0.67 pp over the three-seed no-CGCM mean; operator controls indicate that this gain is not explained by auxiliary.
37. Geometry Without Coordinates: LiDAR Diffusion as a 3D Feature Bridge
Core Problem: Transferring the rich priors of large 2D foundation models to sparse 3D LiDAR remains challenging, as training native 3D foundation models at comparable scale is limited by data and annotation scarcity.
Key Innovation: We introduce a LiDAR-conditioned diffusion model trained on pseudo-labels from off-the-shelf 2D foundation models. Pairwise cosine similarity across modality-specific feature streams reveals a layered organization.
38. Dimensionality Reduction for Hyperspectral Image Classification
Core Problem: This paper addresses the issue of supervised classification in the context of hyperspectral satellite images.
Key Innovation: It deals with two fundamental aspects: dimensionality reduction of data and the selection of appropriate supervised classification techniques. The results highlight that the combination of PCA and RF yields the highest overall accuracy and Kappa coefficient.
39. Beyond Weak Labels: Prompt-Guided Local Refinement for Weakly Supervised Water Segmentation in High-Resolution Multispectral Imagery
Core Problem: High-resolution water mapping supports environmental monitoring and related applications, but accurate pixel-level labels are difficult and costly to produce.
Key Innovation: We propose a two-stage framework for weakly supervised water segmentation in high resolution multispectral imagery. The results indicate that prompt-guided refinement can improve pseudo-label-based water segmentation by targeting local errors that are poorly captured by global training supervision.
40. Semigroup-JEPA: Latent Dynamics Consistency for Zero-Shot Physics Generalization
Core Problem: Joint-Embedding Predictive Architecture (JEPA) world models learn a compact latent representation of the world that supports prediction and planning, but their capability to learn physics and generate physically realistic dynamics remains hitherto untested.
Key Innovation: In this work, we introduce SemiGroup-JEPA (SG-JEPA), which extends the LeWorldModel framework by supplying the parameter governing the physics to the temporal model via action-conditioning and jointly training an encoder and predictor through an autoregressive latent rollout.
41. Control Copy-Paste: Controllable Diffusion-Based Augmentation Method for Remote Sensing Few-Shot Object Detection
Core Problem: Few-shot object detection (FSOD) for optical remote sensing images aims to detect rare objects with only a few annotated bounding boxes.
Key Innovation: Accordingly, we propose Control Copy-Paste, a controllable diffusion-based method to enhance the performance of FSOD by leveraging diverse contextual information. The limited training data makes it difficult to represent the data distribution of realistic remote sensing scenes, which results in the notorious overfitting problem.
42. DCReg: Decoupled Characterization for Efficient Degenerate LiDAR Registration
Core Problem: Existing detect-then-mitigate methods fail to reliably detect, physically interpret, and stabilize this ill-conditioning without corrupting the optimization.
Key Innovation: We introduce DCReg (Decoupled Characterization for Ill-conditioned Registration), establishing a detect-characterize-mitigate paradigm that systematically addresses ill-conditioned registration via three innovations. First, DCReg achieves reliable ill-conditioning detection by employing Schur complement decomposition on the Hessian matrix.
43. Diverse Instance Generation via Diffusion Models for Enhanced Few-Shot Object Detection in Remote Sensing Images
Core Problem: Few-shot object detection (FSOD) aims to detect novel instances with only a limited number of labeled training samples, presenting a challenge that is particularly prominent in numerous remote sensing applications such as endangered species monitoring and disaster assessment.
Key Innovation: To address this issue, we propose a novel framework that can leverage a diffusion model pretrained on large-scale natural images to synthesize diverse remote sensing instances, thereby improving the performance of few-shot object detectors. Existing FSOD methods for remote sensing images (RSIs) have achieved promising progress but remain constrained by the limited diversity of instances.
44. Steering Diffusion Priors with Sparse Observations for High-Resolution Temperature Downscaling
Core Problem: Local heatwave hazard depends on fine-scale air temperature, but ground stations are sparse and reanalysis products such as ERA5 cannot resolve the terrain and land-surface contrasts that shape real heat exposure.
Key Innovation: We present a conditional diffusion emulator for high-resolution 2-m temperature downscaling, conditioned on static geography, a training climatology, exact-time ERA5 temperature, and solar and temporal features, guided at inference by score-based data assimilation (SDA): a differentiable Gaussian observation likelihood steers the diffusion score toward sparse revealed temperature observations without any retraining.
45. Agent-Based ML-LLM Fusion with Self-Optimizing Prompts for Plateau Weather Alerts
Core Problem: Key improvements include a sharp rise in B08 from data source citation (6.5 -more than 8.5), sustained high performance in B10 via physical mechanism explanation, and a peak scientific rigor score of 9.2 in B12 through explicit uncertainty statements.
Key Innovation: To address insufficient contextualization, weak generalization, and poor scenario adaptation in tourism meteorological services, we propose SmartWeatherAgent--a unified three-stage architecture integrating intent recognition, hazard prediction, and reasoning-enhanced generation.
46. SCPT-Based Monotonic Soil Reaction Model for Monopiles in Sand: Calibration with Centrifuge Tests and Assessment against Field Data
Core Problem: This paper presents a soil reaction framework for monotonic lateral analysis of offshore wind monopiles in sand, formulated using input parameters that can be derived from seismic cone penetration test (SCPT) data.
Key Innovation: This paper presents a soil reaction framework for monotonic lateral analysis of offshore wind monopiles in sand, formulated using input parameters that can be derived from seismic cone penetration test (SCPT) data.
47. Extreme storm tides along the northern South China Sea Coast: Tide-Surge decomposition and historical intensification
Core Problem: However, previous studies in this region have largely focused on individual event or short-term simulation, leaving gaps in understanding the long-term spatiotemporal variability of extreme storm tides (ESTs) and the contributions of their components.
Key Innovation: In this study, TC-induced storm surges along the northern SCS coast from 1979 to 2020 were systematically simulated to characterize the spatiotemporal EST variability. The results reveal pronounced spatial heterogeneity in ESTs.
48. Rolling multi-step forecasting of monthly mean sea level using data decomposition-based ensemble prediction models
Core Problem: Monthly mean sea level (MMSL) is increasingly affected by climate change and human activities, resulting in pronounced non-stationarity that poses challenges for forecasting.
Key Innovation: This study aims to fill this gap by developing a unified MMSL forecasting benchmark across eight sea areas, which facilitates comprehensive and fair comparisons of various modeling strategies and data decomposition methods. Among the tested EPMs, the coupling model of time varying filtering based empirical mode decomposition and ENN (TVFEMD-ENN) achieves the best overall accuracy and robustness.
49. A Directional and Multiscale Features-Based Method to Extract Ship Wake in SAR Images
Core Problem: However, the similarity between ship wakes and the complex background of the ocean surface leads to low detection accuracy.
Key Innovation: Herein, we propose a YOLOv8s-OBB-based ship wake detection network YOLO-RealSW. Experiments on the public dataset OpenSARWake show that this method is superior to the existing mainstream methods in terms of accuracy and robustness.
50. Multisource Data Fusion for Snowmelt Runoff Onset Date Estimation in Forested Areas
Core Problem: However, when using SAR data for ROD estimation in forested areas, microwave signals are severely attenuated due to scattering and reflection from branches and tree trunks, making it difficult to capture the state changes of snow beneath the forest canopy.
Key Innovation: Considering the strong correlation between ROD and vegetation phenology, this study proposes a machine learning-driven framework that leverages extreme gradient boosting (XGBoost) to enhance the accuracy of ROD estimation in forested areas through the synergistic fusion of multisource heterogeneous data.
51. DTAC-3DMesh: A 3-D Mesh Modeling Method for Buildings Using ArrayInSAR Point Clouds
Core Problem: Consequently, they often exhibit sparse sampling, incomplete roof and facade coverage, and complex noise.
Key Innovation: To address these limitations, this study proposes DTAC-3DMesh, a method for reconstructing 3-D building meshes from ArrayInSAR point clouds.
52. Land-Oriented Scene Graph Generation for High-Resolution Remote Sensing Imagery: A Specialized Dataset and Semantic-Visual Collaborative Method
Core Problem: However, existing scene graph generation (SGG) methods are difficult to adapt to remote sensing imagery due to the lack of dedicated land-oriented benchmarks, semantic-visual inconsistency, and severe long-tailed relationship distributions.
Key Innovation: To address these issues, this study constructs the first land-oriented remote sensing SGG dataset by integrating and refining land-scene samples from ReCon1M and satellite-based terrain and relationship, containing 19 658 images, 50 object categories, and 45 relationship categories. Experimental results show that, compared with PE-Net, the proposed method improves R@100/mR@100 from 43.59/22.75 to 52.81/36.51 under scene graph.
53. Hyperspectral Satellite-Derived Phytoplankton Diagnostic Pigments in the East China Sea: Spatiotemporal Patterns and Ecological Response to Coastal Fronts
Core Problem: Phytoplankton diagnostic pigments are critical for identifying phytoplankton functional types, yet their retrieval is limited by sparse in situ data and low spectral resolution.
Key Innovation: This study retrieved phytoplankton diagnostic pigments in the East China Sea using hyperspectral PACE ocean color data with 5 nm resolution in the visible range. The model performance was validated against multispectral MODIS-derived results, and its robustness was further verified across independent random and two out-of-distribution generalization scenarios.
54. DFDSCF: A Dual-Frequency and Dynamic Sparse Cross-Path Fusion Network for Hyperspectral Image Classification
Core Problem: Existing models still face challenges in simultaneously capturing fine-grained structural details and discriminative spectral patterns, as well as in enabling sufficiently deep interaction between spatial and spectral modalities, thereby limiting their effectiveness in complex scenes.
Key Innovation: Existing models still face challenges in simultaneously capturing fine-grained structural details and discriminative spectral patterns, as well as in enabling sufficiently deep interaction between spatial and spectral modalities, thereby limiting their effectiveness in complex scenes.
55. PSMamba: Conditional Multiscale Mamba for Remote Sensing Image Pan-Sharpening
Core Problem: Pansharpening combines panchromatic (PAN) and multispectral (MS) images to recover high-resolution MS imagery.
Key Innovation: We instead regard the PAN image as a spatial constraint and propose PSMamba, an asymmetric conditional guidance framework.
56. Automated Semantic Segmentation Label Generation for Large-Scale Wetland Mapping Using Local Gaussian-Weighted Reverse K-Nearest Neighbor
Core Problem: Wetland mapping is a fundamental prerequisite for ecological conservation and management, yet large-scale applications are severely hindered by the difficulty of acquiring high-quality and low-noise training samples.
Key Innovation: To address these challenges, this study proposes a highly automated and reliable framework for training sample generation and optimization in large-scale wetland mapping. Cross-regional generalization experiments conducted across seven ecologically heterogeneous regions demonstrate that the proposed framework yields remarkable performance improvements against initial labels produced by random forest, with the overall accuracy.
57. Adaptive Point Selective Deformable Attention and Multiscale 3-D Convolution Dynamic Fusion for Hyperspectral Band Selection
Core Problem: Due to treating all features equally or relying on fixed single-scale convolutional kernels, most current methods fail to capture complex structures or subtle spectral variations within images, resulting in suboptimal performance.
Key Innovation: Due to treating all features equally or relying on fixed single-scale convolutional kernels, most current methods fail to capture complex structures or subtle spectral variations within images, resulting in suboptimal performance. Experimental results demonstrate that the proposed method outperforms other state-of-the-art approaches.
58. RadarDiff: A Conditional Diffusion Model With Time-Aware Context and Mixed Attention for Radar Echo Extrapolation
Core Problem: Accurate radar echo extrapolation is essential for precipitation nowcasting.
Key Innovation: Existing deterministic models suffer from regression to the mean and are prone to oversmoothing, leading to distorted representations of strong convective cores. To address these issues, this article proposes a conditional diffusion model named RadarDiff, which achieves a better tradeoff among forecast performance, image quality, and computational efficiency through the synergistic design of three core modules.
59. Enhancing Target and Rectifying Clutter: A Structure-Guided Frequency Rectification Network for Infrared Small-Target Detection
Core Problem: Infrared small-target detection (IRSTD) remains challenging under low signal-to-noise ratio conditions, where weak target responses are easily submerged by complex background clutter.
Key Innovation: Accordingly, we propose a structure-guided frequency rectification network (SGFR-Net) for IRSTD. Extensive experiments on SIRST, NUDT-SIRST, and IRSTD-1 K demonstrate that SGFR-Net achieves strong segmentation accuracy and robust detection performance under complex backgrounds.
60. End-to-End Deep Reconstruction and Clustering of Satellite Image Time Series for Crop Group Mapping
Core Problem: On one hand, training data for such applications may be scarce, especially when relatively wide scenes are considered, thus prompting the need for unsupervised approaches.
Key Innovation: In this article, we propose a neural architecture that performs gap filling and clustering simultaneously in an end-to-end framework for crop group classification. Experimental results on two datasets suggest the effectiveness of the proposed method as well as the relevance of all of its architectural components.
61. Improving UAV-Derived Nearshore Bathymetry Using Geospatial Machine Learning
Core Problem: However, nearshore regions remain undersurveyed due to dynamic environmental conditions and the limitations of traditional multibeam echosounders (MBES).
Key Innovation: This study presents a framework that integrates high-resolution uncrewed aerial vehicle (UAV) multispectral imagery with geospatial machine learning to address spatial variability in reflectance-depth relationships and to mitigate performance overestimation by explicitly accounting for spatial autocorrelation.
62. Itoh-Constrained Hybrid Graph Learning With Sparse Supervision for InSAR Phase Unwrapping
Core Problem: Phase unwrapping (PU) is a fundamental step in synthetic aperture radar interferometry (InSAR), yet it remains an inherently ill-posed problem due to the 2\pi ambiguity of the interferometric phase.
Key Innovation: Existing deep learning-based PU methods typically rely on dense pixelwise supervision using absolute phase values, which introduces a fundamental mismatch with the intrinsic structure of PU and often leads to poor generalization. Extensive experiments demonstrate that the proposed method consistently outperforms dense supervision and other baseline methods under both uniform and nonuniform sampling conditions.
63. A Two-Step Machine Learning Framework for High-Resolution Reconstruction of Regional GNSS-Derived PWV Grids Using Meteorological Auxiliary Data
Core Problem: PWV retrieved from the Global Navigation Satellite System (GNSS) offers high temporal resolution and all-weather observation capability; however, the sparse and uneven distribution of GNSS stations limits its direct application in constructing continuous high-resolution moisture fields.
Key Innovation: To address this limitation, this study integrates discrete GNSS-PWV observations with high-resolution rapid refresh meteorological information and develops a spatial reconstruction framework to generate PWV grids with a resolution of 0.01^{\circ }× 0.01^{\circ }, enabling high-resolution reconstruction from sparse station observations to a continuous high-resolution water vapor field.
64. M2GCN: A Multiadjacency Mamba-Enhanced Graph Convolutional Network for Hyperspectral Image Classification
Core Problem: However, existing methods based on GCN primarily rely on single adjacent relationship modeling and do not fully utilize the edge relationships between superpixels, which limits their classification performance in complex terrain scenarios.
Key Innovation: However, existing methods based on GCN primarily rely on single adjacent relationship modeling and do not fully utilize the edge relationships between superpixels, which limits their classification performance in complex terrain scenarios. Extensive experiments demonstrate that, compared to the state-of-the-art method, M2GCN achieves classification performance improvements of 1.63%, 1.40%, and 1.29% on the Indian Pines.
65. Mitigating False Positives in Complex Spectral Scenarios: A Mutual Information Guided Wavelet Disentanglement Framework for Remote Sensing Change Detection
Core Problem: To systematically address these challenges, we propose ACNet, a graph-enhanced adaptive wavelet transform network with cross-layer knowledge injection and mutual information regularization.
Key Innovation: To systematically address these challenges, we propose ACNet, a graph-enhanced adaptive wavelet transform network with cross-layer knowledge injection and mutual information regularization. Extensive experiments on five widely-used benchmark datasets demonstrate that ACNet consistently outperforms ten state-of-the-art methods, capable of detecting both subtle and large-scale changes while substantially mitigating false.
66. COWVR Wind Direction Performance for One-Look Versus Two-Look Viewing Geometry
Core Problem: The Compact Ocean Wind Vector Radiometer (COWVR) instrument is a technological demonstration of a compact, low-cost polarimetric microwave radiometer for measuring wind speed and direction over the ocean surface.
Key Innovation: COWVR’s internal calibration system enables Earth observations over the entire scan and a two-look wind direction retrieval. When we compare COWVR’s two-look wind directions with retrievals that use only one of COWVR’s looks, we find that certain relative wind directions are prone to error in one-look retrievals.
67. ISGM: An Illumination-Aware Semantic-Guided Mamba Network for RGB-Infrared Vehicle Detection in UAV Imagery
Core Problem: RGB-infrared (RGB-IR) vehicle detection in uncrewed aerial vehicle (UAV) imagery is essential for applications, such as traffic monitoring and object tracking.
Key Innovation: To alleviate these issues, we propose an Illumination-aware Semantic-Guided Mamba (ISGM) network. Extensive experiments on the DroneVehicle and VEDAI datasets demonstrate that ISGM outperforms state-of-the-art methods in detection performance while achieving a favorable accuracy-efficiency balance among comparable methods.
68. Decoupling Clutter Suppression and Target Localization: A Two-Stage Framework for Robust Radar Detection
Core Problem: Complex sea clutter remains a major challenge for radar target detection because of its non-Gaussian statistics, heavy-tailed fluctuations, and pronounced nonstationary behavior, which severely degrade the reliability of weak-target detection.
Key Innovation: Existing deep learning methods usually perform clutter suppression and target localization within a single-stage framework, where the two objectives are optimized simultaneously in a shared feature space.
69. A 19-year record of atmospheric sulphur dioxide (SO₂) derived from IASI measurements
Core Problem: In this paper, we present a 2007-2026 record of twice-daily global SO₂ vertical column abundances and SO₂ plume altitudes derived from measurements by the three Infrared Atmospheric Sounding Interferometer (IASI) instruments onboard the Metop platforms.
Key Innovation: In this paper, we present a 2007-2026 record of twice-daily global SO₂ vertical column abundances and SO₂ plume altitudes derived from measurements by the three Infrared Atmospheric Sounding Interferometer (IASI) instruments onboard the Metop platforms.
70. Circum-Arctic Sediment PROvenance Database (CASPROD): a database of mineralogy and geochemistry for the Circum-Arctic surface sediments
Core Problem: Abstract.
Key Innovation: While surface sediments in this semi-enclosed basin integrate complex signals from diverse Eurasian and North American source regions, disentangling these provenance signatures requires a robust, multi-proxy framework that has historically been hampered by fragmented, heterogeneous datasets.
71. Observations from a 94-GHz spectral polarimetric vertically-pointing radar at Villum research station during the CLAVIER Clean Cloud campaign
Core Problem: Abstract.
Key Innovation: Peak-resolved spectral parameters are subsequently used within a hydrometeor classification framework implemented in pyLARDA enabling the identification of liquid droplets, pristine ice crystals, aggregates and rimed particles.
72. The ATMOSFER campaign: synergistic observations for the preparation of the ESA FORUM mission
Core Problem: Abstract.
Key Innovation: The balloon-borne ATMOSFER campaign, supported by CNES, was held in Kiruna in June 2024. This campaign involved lightweight hygrometers (Pico-Light H₂O, NOAA GML frost point hygrometer, EN-SCI Cryogenic frost point hygrometer), optical particle counters and ice crystal imager (POPS, LOAC v1.5, LPC and NIXE-B), ozone sondes and the airborne demonstrator of the future ESA FORUM mission developed by CNR-INO (FIRMOS-B instrument).
73. Three-dimensional geological modeling based on dual-task stratigraphy-aware attention networks (Geo-SAN v1.0)
Core Problem: The current three-dimensional (3D) geological implicit modelling methods are mainly based on interpolation methods, such as Kriging and radial basis functions (RBFs), which struggle to capture the nonlinear characteristics of complex geological structures and are limited in their capacity to integrate multi-source modeling data.
Key Innovation: To overcome these limitations, we proposed a 3D geological modelling framework, Geo-SAN, which consists of a dual-task stratigraphy-aware attention network. A case study at the Lingnian-Ningping region of Guangxi Zhuang Autonomous Region (GZAR), China, demonstrates that the proposed Geo-SAN framework, with an accuracy of 92.1% in lithological classification and a coefficient of determination (R²) of 0.96 in predicting the.
74. Historical evolution of snowpack capacity to buffer rain-on-snow runoff in a large Columbia River headwaters basin
Core Problem: However, a given sized rain-on-snow (ROS) event can yield outcomes ranging from flooding to no runoff, depending partly on the snowpack’s antecedent cold content and capillary retention forces.
Key Innovation: We use ERA-5 Land data to force a snowpack model that tracks the layer-by-layer development of heat, mass, and structural framework of the snowpack throughout the snow season. The core five weeks of mid-winter showed no trending change of LWbc, and in fact demonstrated an increase in cold content over the 72 years.
75. Hydroclimatic-driven pulse enrichment amplifies riverine microplastic export in developing countries
Core Problem: Rivers are key conduits in the global microplastic cycle, but how much they export to ` ocean and what controls this export remain unresolved.
Key Innovation: Here, we present a global daily dataset of riverine microplastic concentrations and exports and find that rivers delivered far more microplastics to the ocean than most previous estimates suggest.
76. Adaptive Neural Network Approaches in Remote Sensing Imagery: A Systematic Review
Core Problem: However, these data are often affected by sensor differences, spatial and temporal heterogeneity, missing observations, irregular sampling, noise, and non-stationary environmental processes.
Key Innovation: However, these data are often affected by sensor differences, spatial and temporal heterogeneity, missing observations, irregular sampling, noise, and non-stationary environmental processes. The findings show that adaptive neural networks have gained increasing attention after 2022 and are mainly applied to image-centered remote sensing tasks, including classification, mapping, object detection, segmentation, enhancement, and.
77. Extending Multidimensional Rao’s Quadratic Entropy to Optical-Radar Lava-Flow Mapping Using Sentinel-1 and Sentinel-2: Evidence from the 2021 La Palma Eruption
Core Problem: Optical, radar, thermal infrared, and night-time radiance datasets were evaluated within a common change-detection framework implemented in Google Earth Engine.
Key Innovation: This study assessed direct spectral, classic Rao’s quadratic entropy (RaoQ), and multidimensional RaoQ approaches using satellite observations acquired before and after the eruption. Among the direct spectral approaches, the NHI_SWIR index achieved the highest overall classification performance.
78. MTCPNet: A Mamba-Based Registration Network with Tri-Branch Consistency Projection for SAR-Visible Image Registration
Core Problem: Visible and synthetic aperture radar (SAR) images exhibit substantial nonlinear radiometric differences and geometric deformations due to their fundamentally different imaging mechanisms, making high-precision registration between the two modalities a long-standing challenge in remote sensing image processing.
Key Innovation: Existing deep learning-based cross-modal registration methods mostly adopt purely convolutional architectures or hybrid convolution-Transformer frameworks, which struggle to achieve a favorable trade-off between long-range dependency modeling and computational efficiency.
79. Assessing the Contributions and Thresholds of Multi-Source Remote Sensing Informatiuon in Vegetation Aboveground Biomass Estimation Under Optical Saturation
Core Problem: However, optical-remote-sensing-based AGB estimation is prone to saturation effects.
Key Innovation: Accordingly, this study used Dongzhai National Nature Reserve in Luoshan County, Henan Province, China, as the study area and 90 field-measured AGB values as the target variable; integrated GEDI L2A/L2B, Landsat 9, Sentinel-2, GF-2, GF-5B hyperspectral, Sentinel-1, topographic, and climatic data; constructed four feature sets comprising optical information Sp, structural information St, microwave scattering information Se.
80. ZOS-Net: A Lightweight RGB-T Object Detection Network with Cross-Modal Relation Enhancement and Selective Target Awareness
Core Problem: Existing multimodal detection methods typically rely on complex attention structures or heavyweight fusion modules, making it difficult to balance detection accuracy, model lightweightness, and deployment efficiency; moreover, modality noise and redundant background information are prone to joint propagation during shallow fusion, weakening the responses of small and weak objects.
Key Innovation: Specifically, ZOS-Net introduces a Cross-Modal Fine-Grained Gated Fusion module, termed ZCGF, at the shallow P3 stage to enhance reliable complementary information from visible textures and infrared thermal responses; an Object-Aware Relation Enhancement module, termed OAGR, is introduced at the semantic bridging stage from P4 to P3 to generate object-relation priors; and a Selective Relation-Guided Target Perception.
81. Scattering-Semantic Collaborative Learning via Asymmetric Dual-Branch DINO Network for Inshore SAR Ship Detection
Core Problem: Inshore synthetic aperture radar (SAR) ship detection remains challenging because strong coastal clutter, speckle noise, and degraded target responses frequently result in false alarms and missed detections of small vessels.
Key Innovation: Moreover, local scattering-related responses and high-level semantic information exhibit different characteristics across network stages, making it difficult for a unified feature-learning framework to fully exploit their complementarity. Experiments on the SSDD and HRSID inshore subsets demonstrate that, compared with the baseline DINO, S-DINO improves mAP@50 by 5.6 and 11.5 percentage points and F1-score by 11.4 and 8.6.
82. Properties and Temporal Evolution of an Elevated Arctic Liquid Fog by Lidar
Core Problem: In this work, we present a case study mainly from Raman lidar observations and radiosonde data for a site in the European Arctic to analyse how aerosol grows into a purely water-containing fog.
Key Innovation: In this work, we present a case study mainly from Raman lidar observations and radiosonde data for a site in the European Arctic to analyse how aerosol grows into a purely water-containing fog. We find predominantly accumulation-size particles before and after the fog event.
83. Sentinel-2 Forel-Ule Index as a Proxy for Ecological Status in Reservoirs: A Case Study in Southern Portugal
Core Problem: Water color is an important optical proxy for trophic status and water quality, but its integration into regulatory assessment frameworks is still limited.
Key Innovation: Water color is an important optical proxy for trophic status and water quality, but its integration into regulatory assessment frameworks is still limited. The results showed that the values on the FUI scale (which ranges from 1 to 21) fell, for the most part, between 12 and 18 and with marked spatial and seasonal contrasts, particularly between more transparent reservoirs and persistently turbid ones, probably eutrophicated.
84. Plastic Greenhouse Extraction Based on Index-Guided Three-Stage Large-Factor Remote Sensing Image Super-Resolution Reconstruction
Core Problem: Fine-scale, large-area plastic greenhouse (PG) mapping generally relies on costly high-resolution imagery with limited spatial coverage.
Key Innovation: To address these challenges, we propose a three-stage 12× SR framework guided by a spatially enhanced Agricultural Plastic Greenhouse Index (APGI). Experiments conducted in Weifang, China, and Almería, Spain, demonstrate that progressive reconstruction and APGI guidance improve PG boundary continuity, reduce extraction errors, and enhance downstream segmentation accuracy.
85. Crop Yield Estimation with MODIS Derived Normalized Difference Vegetation Index and Comparative Study on Crop Yield Prediction Among Linear Regression, Random Forest and Gradient Boosting as Well as CatBoost
Core Problem: Although many prior studies address crop-yield prediction with linear regression, random forest, gradient boosting, and related methods, a complementary, aggregate-level verification method for predicted crop yield has rarely been proposed.
Key Innovation: Although many prior studies address crop-yield prediction with linear regression, random forest, gradient boosting, and related methods, a complementary, aggregate-level verification method for predicted crop yield has rarely been proposed.
86. A Spatial Prior-Guided Feature Enhancement and Multi-Branch Complementary Learning Framework for Small Ship Detection in SAR Images
Core Problem: Ship detection in Synthetic Aperture Radar (SAR) imagery is essential for maritime surveillance and situational awareness.
Key Innovation: To alleviate this dilemma, a Spatial Prior-guided feature enhancement and Multi-branch Complementary learning framework is proposed, termed SPMC, for small ship detection in SAR images. For extremely small ships with very limited image coverage, SPMC improves the baseline YOLOv11 detector by 3.25%/0.99% in AP50/AP0.5:0.95 on LS-SSDD and by 1.28%/1.21% on HRSID, demonstrating its effectiveness in challenging SAR small ship.
87. A Comprehensive Evaluation of Deep Learning-Based Image Super-Resolution for GCP Chip Matching Using High-Resolution Satellite Imagery
Core Problem: High-resolution (HR) Ground Control Point (GCP) chips are essential for accurate satellite image registration, yet their generation commonly relies on aerial imagery, which is often limited by data availability and update frequency.
Key Innovation: To improve the usability of satellite-derived GCP chips as an alternative, this study systematically investigates the influence of deep learning-based image super-resolution (SR) on GCP chip template matching under varying input spatial resolution conditions. The experimental results show that SR is more effective for input GCP chips with limited spatial detail, with the largest improvements obtained under the ×4 downsampling.
88. Hierarchical Sparsity-Guided DeepLassoNet: Interpretable PolSAR Feature Selection for Typical Land-Cover Classification in the Hunshandake Sandy Land
Core Problem: Given the need for reliable land-cover monitoring in such a fragile and dynamically changing ecosystem, spaceborne polarimetric synthetic aperture radar (PolSAR) is adopted for typical land-cover classification to support grassland ecological monitoring and desertification control.
Key Innovation: To address this issue, we propose a DeepLassoNet-based feature selection method for typical land-cover classification in the Hunshandake Sandy Land. Experimental results demonstrate that the proposed method can effectively select discriminative PolSAR features, reduce feature redundancy interference, and improve the classification accuracy and robustness in complex sandy land-grassland transition scenarios.
89. Deep Learning-Based Classification of Plunging Breaker Conditions Using Simulation Radar HRRP Sea-Surface Scattering Data
Core Problem: Electromagnetic scattering from plunging breaking waves generates strong sea-surface radar returns that degrade radar-based maritime surveillance and target detection performance.
Key Innovation: This study develops a deep learning framework for automatic classification of simulated plunging-breaker scattering conditions using high-range-resolution profile (HRRP) data. The ANN achieves an overall classification accuracy of 96%, compared with 91% for the 1D ResNet CNN under the simulated dataset and adopted model configurations.
90. Semantic-Driven Adversarial Reconstruction Learning for Open-Set Recognition in Remote Sensing Imagery
Core Problem: Open-Set Recognition (OSR) in Remote Sensing Scene Images (RSSIs) is severely hindered by complex backgrounds, which obscure the generative failures of unknown classes in traditional reconstruction-based methods.
Key Innovation: To address this, we propose a Semantic-Driven Adversarial Reconstruction (SDAR) framework that shifts the OSR paradigm to targeted semantic verification. A two-stage training protocol is utilized: the encoder is first trained via classification to extract high-level semantic features, after which the decoder is unfrozen for joint training to achieve semantic-level reconstruction.
91. Biopolymer-modified soil-rock mixtures for open-pit mine waste-dump slopes: investigation of mechanical behavior and microscopic mechanisms
Core Problem: The reshaping of soil-rock mixtures (S-RM) on slopes within the secondary stripping zones of open-pit mines constrains both resource recovery and production safety.
Key Innovation: In this study, calcium lignosulfonate (CLS) and xanthan gum (XG) were used to enhance the mechanical performance of S-RM. Results show that at CLS and XG contents of 3% and 1.
92. Dynamic properties as well as long-term stability of palm fiber reinforced soil
Core Problem: Long-term vehicular loading induces cracking in embankment soil, which can progressively develop into structural failure of the soil.
Key Innovation: To address this issue and improve the embankment soil’s long-term stability, palm fiber-lime-soil (FLS) treatment was adopted. Finally, scanning electron microscopy (SEM) tests were performed to reveal the synergistic mechanism by which palm fiber reinforcement and lime stabilization.
93. Unraveling multi-scale drivers of Chl-a dynamics via interpretable machine learning: a case study of Shahu Lake
Core Problem: Aquatic eutrophication exhibits pronounced temporal variability, yet conventional global interpretation frameworks often fail to capture short-term ecological responses and obscure environmental drivers that dominate only during specific ecological periods.
Key Innovation: Aquatic eutrophication exhibits pronounced temporal variability, yet conventional global interpretation frameworks often fail to capture short-term ecological responses and obscure environmental drivers that dominate only during specific ecological periods.
94. Lithological and mesoclimatic controls on denudation rates in transform margin crystalline massifs: Insights from cosmogenic 10Be and 26Al in semi-arid Brazil
Core Problem: The landscapes of the transform margin of semi-arid Brazil are characterized by planation surfaces dotted with varied residual reliefs.
Key Innovation: In this context, denudation rates were measured in crystalline massifs with lithological and climatic contrasts, evaluating the role of differential erosion and mesoclimatic gradients in their morphogenesis. The results reveal a denudational contrast between dry (leeward) and humid (windward) slopes (apparent rates of 18.3 ± 1.5 and 13.7 ± 1.2 m/Myr) that becomes sharper and statistically significant after correction for.
95. Integrating PS-InSAR and SDG-aligned indicators for risk-informed urban prioritization in Ravenna, Italy
Core Problem: Sustainable urban management increasingly requires sub-municipal tools to identify where risk, degradation, and limited adaptive capacity most urgently require intervention.
Key Innovation: This study presents a transparent, open-data framework for identifying intervention priorities. Results indicate a mean subsidence rate of −2.5 mm/year, with over 1500 buildings beyond −5 mm/year.
96. Physics-informed vessel trajectory prediction via semantic-kinematic coupling for intelligent waterway oversight
Core Problem: First, to address the challenge of interpreting asynchronous and fragmented data in dense traffic scenarios, we establish a state representation pipeline.
Key Innovation: To overcome these limitations and enhance navigational safety, this paper proposes a novel physics-informed framework that leverages semantic-kinematic coupling for intent-driven trajectory prediction. Evaluations on real-world inland AIS datasets demonstrate that the framework reduces long-term displacement errors and physical violations compared to baseline models.
97. Identifying high-risk lines for power system cascading failures: A multi dimensional feature enhanced two-stage learning approach for safety and resilience
Core Problem: Complex operating conditions and external disturbances increase the risk of cascading failures in power systems, threatening system safety and service continuity.
Key Innovation: Critical line identification is reformulated as a continuous risk regression problem, and the peak-over-threshold method from extreme value theory is introduced to statistically define critical line thresholds. Case studies on the IEEE 39-bus system demonstrate a 7.10% improvement in identification accuracy and a 12.4% reduction in combined defense and blackout costs, while tests on the IEEE 118-bus system further verify.
98. Four decades of Landsat-based river turbidity reconstruction reveal spatially heterogeneous declines across U.S. rivers
Core Problem: Despite prior large-scale optical water-quality studies of U.S. rivers, systematic, multi-decadal analyses integrating Landsat surface reflectance, machine-learning retrieval, and ecoregion-scale driver attribution have remained limited.
Key Innovation: This study developed and validated a long-term Landsat-based turbidity reconstruction framework for major rivers of the conterminous United States from 1984 to 2025.
99. PELNet: A general physics-embedded learning framework for polarimetric target decomposition
Core Problem: Traditional model-driven methods often suffer from rigid power allocation, sensitivity to noise, and limited capability in representing complex scattering scenes.
Key Innovation: Building on these observations, we investigate the integration of physical priors into deep neural networks and propose PELNet, a physics-embedded end-to-end framework for polarimetric target decomposition. Experiments on multi-band PolSAR datasets across diverse terrain types demonstrate that PELNet achieves superior accuracy and interpretability compared to traditional methods, providing a new perspective for PolSAR image.
100. Synergizing motion and depth: A distilled dual-branch network for moving infrared small target detection
Core Problem: Moving infrared small target detection (MISTD) is challenging because targets are weak and backgrounds are complex and dynamic.
Key Innovation: To address the limitations, we propose a novel MISTD framework that synergizes Motion and Depth, termed MoDe, to suppress background responses in both the spatial and temporal domains. Existing approaches have achieved notable progress in spatiotemporal feature representation for MISTD, but often suffer from limited robustness in cluttered scenes due to insufficient background suppression.
101. GICL-Net: A graph-interactive collaborative learning network for joint classification of HSI and LiDAR
Core Problem: However, determining the intrinsic correlation characteristics between HSI channels and fully exploiting the complementary characteristics of HSI and LiDAR to improve classification accuracy remains a challenging issue.
Key Innovation: In this paper, we propose a graph-interactive collaborative learning network (GICL-Net) for joint classification of HSI and LiDAR, which consists of a multi-scale spectral-spatial graph feature extractor (MSGE) and a cross-modality adjacency matrix fusion (CAMF) module. The CAMF module leverages the fused adjacency matrix to establish a cross-modal graph interaction mechanism, enabling the spectral features and the spatial.
102. Utilizing thermal infrared spectroscopy data to map lithology and mineral alteration in geothermally active areas
Core Problem: The potential of thermal infrared (TIR: 8 to 13 μm) imaging spectroscopy for geothermal mineral mapping remains largely under-explored.
Key Innovation: Here we present the first ever lithological and mineralogical maps of Parco Naturalistico delle Biancane (PNB), part of the Larderello geothermal field, derived from TIR imaging spectroscopy data collected by the Hyperspectral Thermal Emission Spectrometer (HyTES). By integrating TIR-derived surface mineralogy with surface temperature, we demonstrate a spatial correlation, linking mineralogical alteration to geothermal.
103. Structural stability of vegetation greening across India from multi-season MODIS EVI: a time-frequency earth observation framework
Core Problem: Large-scale vegetation greening has been widely reported across India; however, national assessments have largely relied on monotonic trend metrics without explicitly evaluating seasonal variability, spatial heterogeneity, or the structural persistence of vegetation dynamics.
Key Innovation: This study presents a physiography-resolved, multi-season assessment of vegetation change using a 24-year (2001-2024) MODIS 16-day Enhanced Vegetation Index (EVI; 250 m) dataset spanning 14 physiographic subdivisions.
104. SegRail: A Framework for semantic segmentation of long-distance railway MLS data via global priors and local geometry integration
Core Problem: To address these challenges, we propose SegRail, a dedicated framework for semantic segmentation of long-distance railway MLS data that explicitly integrates global priors and local geometry.
Key Innovation: To address these challenges, we propose SegRail, a dedicated framework for semantic segmentation of long-distance railway MLS data that explicitly integrates global priors and local geometry. Extensive experiments on five railway datasets demonstrate the effectiveness of SegRail, which achieves the highest mIoU than the compared baseline methods under multiple runs.
105. Soil salinity inversion in croplands of the Yellow River Delta: a feature-enhanced stacking framework integrating multi-source remote sensing and environmental covariates
Core Problem: Accurate characterization of soil salinity spatial distribution is essential for coastal agroecosystem management; however, effectively representing heterogeneous multi-source features and improving model generalization remain challenging in complex coastal environments.
Key Innovation: This study was conducted in Guangrao County, Yellow River Delta, and developed a feature-enhanced Stacking ensemble framework for soil salinity inversion by integrating Sentinel-1 synthetic aperture radar (SAR) data, Sentinel-2 multispectral data, and multi-source environmental covariates. The results showed that distance to the sea, salinity index 1 (SI1), and soil sand content consistently exhibited high importance under.
106. Stokes number regulation on inertial deposition and transport of snow particles over gable roofs
Core Problem: The transport and deposition behavior of snow particles in air-snow two-phase flows is strongly modulated by particle inertia, which is quantitatively characterized by the Stokes number (St).
Key Innovation: To reveal the regulatory mechanism of St on snow distribution over gable roofs, this study adopts an Eulerian-Eulerian two-phase flow model and varies the snow particle diameter to adjust St (ranging from 0.02 to 4.14).
107. Observed and computed performance of mechanised tunnelling in the historical centre of Rome
Core Problem: The project of the Line C of the Rome underground involved the use of two earth-pressure balance (EPB) tunnelling boring machines (TBMs) to excavate the twin tunnels in the city centre, having the priority of minimising potentially detrimental interferences with existing monuments and historical buildings of invaluable historical value.
Key Innovation: This gave rise to a massive monitoring program to evaluate the effects of the mechanised tunnelling on the nearby monuments, providing the opportunity for field investigation of ground response. This paper presents and discusses the monitoring results collected from a greenfield test site located close to the Basilica of San Giovanni and instrumented with extensive surface and subsurface instrumentation.
108. Duration-conditioned temporal profiles and synchronization networks reveal regional organization of midlatitude precipitation events
Core Problem: Short-duration precipitation events are important drivers of flash floods and urban pluvial flooding, yet broad-scale comparisons of their within-event temporal structure and spatial coordination remain limited.
Key Innovation: We analyse quality-controlled hourly rain-gauge observations from the contiguous United States, selected European countries and Japan using a consistent event-based framework. Short-duration events are widely represented, whereas longer-duration events show more localized areas of high station-level fraction.
109. CO₂ migration and storage in layered saline aquifers: effects of vertical high-permeability channel width
Core Problem: However, multiple low-permeability cap rocks and localized high-permeability channels can strongly influence CO₂ migration pathways and storage safety.
Key Innovation: This study addresses the limited understanding of the coupled effects of high-permeability channels on CO₂ migration and multiple trapping mechanisms in multi-layer cap rock conditions. The results show that channel width has a coupled influence on CO₂ migration mechanisms: a 50 m wide channel accelerates pressure propagation and promotes horizontal CO₂ migration (before entering the channel), while a 10 m wide channel.
110. MEDAL: a mechanism-embedded dual-physics attention learning framework for snowmelt-driven runoff prediction in cold-region basins
Core Problem: This study proposes a Mechanism-Embedded Dual-Physics Attention Learning (MEDAL) framework for daily runoff prediction in snowmelt-driven cold-region basins.
Key Innovation: This study proposes a Mechanism-Embedded Dual-Physics Attention Learning (MEDAL) framework for daily runoff prediction in snowmelt-driven cold-region basins. MEDAL achieved the best performance across all eight evaluation metrics in the five study basins.
111. A DEM framework for shear degradation and failure transition of rock joints with finite-thickness weathered joint-wall zones
Core Problem: However, how the degradation degree of this zone affects joint shear resistance, fracture evolution, and failure localization remains insufficiently understood.
Key Innovation: To address this gap, this study develops a strength-weakening-factor-based discrete element method (DEM) framework by incorporating degradation of bond radius and bond strengths into the inherent-microcrack flat-joint contact model, together with spatially selective weakening within a finite-thickness joint-wall zone.
112. Evaluation of severe reinforced concrete building collapses in the context of structural-soil characteristics during the February 6, 2023 Kahramanmaraş earthquakes
Core Problem: The February 6, 2023 Kahramanmaraş Earthquake doublet represents an unusual case in global earthquake history, with two major events of Mw more than 7.0 occurring on the same day in nearby regions.
Key Innovation: In this study, 119 collapsed reinforced concrete buildings located in the most heavily damaged areas of Kahramanmaraş city center were examined in terms of their structural, architectural, and geometric characteristics. The results highlight that the widespread destruction resulted from the combined effects of structural deficiencies, local ground conditions, and soil-structure interaction.
113. Assessment of advanced soil models for seismic soil-structure interaction based on comparisons between centrifuge tests and numerical simulations
Core Problem: Modeling the dynamic behavior of soil-structure interaction is challenging in geotechnical earthquake engineering.
Key Innovation: In this study, the seismic response of a soil-pile-superstructure system is systematically evaluated to clarify how three advanced soil constitutive models - SANISAND and two hypoplastic models - influence seismic behavior. The numerical results for all indicators are validated against centrifuge test data reported in the literature.
114. Seismic-wave-based estimation of anisotropic drained stiffnesses in saturated cohesionless soils: An enhanced analytical approach
Core Problem: Seismic wave measurements provide an efficient means for evaluating small-strain undrained elastic properties in situ; however, practical methods for converting these dynamically derived parameters into drained anisotropic stiffnesses remain limited.
Key Innovation: This study presents an enhanced analytical approach for estimating anisotropic drained stiffnesses from undrained elastic parameters inferred from seismic wave velocities. Predicted drained Young's moduli showed strong agreement with measured values and substantially improved accuracy compared with the previously published approach.
115. Instability analysis and early warning of EPB shield tunnel face induced by muck imbalance
Core Problem: The soil-rock ratio, screw conveyor speed, and tunnelling speed governed muck balance and settlement response, whereas cutterhead speed had a limited influence.
Key Innovation: This study integrated DEM and FEM-DEM simulations, real-time muck monitoring, settlement prediction, and field evaluation to establish a mechanism-informed warning framework. Application to Jinan Metro Line 6 showed that the framework captured changes in settlement risk and supported timely parameter adjustment.
116. Predicting future deformation of frozen soil using the spherical template indenter and the temperature-time analogy method
Core Problem: Rapid determination of long-term strength and deformation of frozen soil remains a fundamental challenge in the design and construction of foundations in cryospheric regions.
Key Innovation: In this study, the spherical template indentation test was employed to rapidly characterize the long-term strength and deformation behavior of frozen soil in Northeast China.
117. Impact of rock discontinuity geometry on uniaxial compressive strength: Integrating the finite-discrete element method with data-augmented interpretable machine learning
Core Problem: Understanding how discontinuity geometric parameters affect the uniaxial compressive strength of complex rock masses (UCSRM) is critical for rock engineering, but traditional methods face challenges in multiparameter analysis due to high costs and inefficiency.
Key Innovation: This study introduces an integrated framework of finite-discrete element method (FDEM), Mixup data augmentation, and interpretable machine learning to predict UCSRM. Expanding the dataset with Mixup, the gradient boosting regressor model achieves a notable accuracy improvement, with the coefficient of determination increasing from 0.813 to 0.899.
118. Dynamic properties of deep rocks under different seepage gradients: Experimental investigation and theoretical modeling
Core Problem: Deep rock engineering projects are frequently conducted in water-rich regions characterized by elevated geo-stress, differential pore water pressure, and frequent dynamic disturbance.
Key Innovation: In this study, a modified triaxial split Hopkinson pressure bar (SHPB) system integrated with hydraulic loading technique was utilized to apply simultaneous differential water pressure and confining pressure. The results demonstrated that increasing the water pressure difference between the two ends of the specimen accelerated the development of internal fractures, enhancing intrinsic permeability and exacerbating.
119. Tensile damage behavior of water-saturated amphibolite at different clay contents: Bridging the micro and mesoscale
Core Problem: Understanding the tensile rupture behavior of rocks with varying clay contents is fundamental for assessing engineering stability.
Key Innovation: This study conducted Brazilian splitting tests on water-saturated amphibolite with different clay contents, investigating the microscopic fracture and damage characteristics using acoustic emission (AE), 3D laser scanning, and scanning electron microscopy (SEM) technologies. The results indicated that varying clay contents resulted in contrasting trends in the splitting mechanical properties of water-saturated amphibolite.
120. Water-induced shear strength deterioration of sandstone-mudstone interface: Experimental and theoretical approaches
Core Problem: Sandstone-mudstone interbedded rock masses are widely distributed in reservoir areas of southwest China.
Key Innovation: This study examined the macroscopic shear performance and microscopic deterioration mechanism of the sandstone-mudstone interface subjected to cyclic drying-wetting treatments via direct shear tests, mineral composition analysis, and microstructure observation. The test findings indicated that both the peak shear strength and the friction angle of the interface decreased in a negative exponential manner as the number of.
121. An Activity-Based Inversion Framework for Estimating Sectoral Anthropogenic CO₂ Emissions Using Multi-Pollutant Satellite Observations
Core Problem: Posterior simulations show improved agreement with Greenhouse Gases Observing Satellite (GOSAT) CO₂ dry column mixing ratios, but gains are limited where CO model-observation mismatches drive large residential emission adjustments, highlighting the need to reduce and better characterize CO-related errors.
Key Innovation: This study develops an activity-based inversion framework to quantify sectoral carbon dioxide (CO₂) emissions using observations of co-emitted air pollutants. Results show that nitrogen dioxide (NO 2) informs energy and transportation adjustments, while SO₂ and CO constrain industrial and residential adjustments.
122. Self-Organization Shapes Divergent Water Use Strategies of Shrubs for Drought Resilience in Drylands
Core Problem: How the coordination between root water uptake and leaf physiological traits drives water use strategies of self-organized shrubs for adapting to drought remains poorly understood.
Key Innovation: Under prolonged drought, dominant shrubs self-organize into two typical spatial configurations: scattered with separated individuals and clumped with clustered individuals. Scattered shrubs showed tight stomatal regulation with high midday leaf water potential and intrinsic water use efficiency (iWUE) during dry season, whereas clumped shrubs exhibited relaxed stomatal regulation, with declines in midday leaf water potential.
123. Physics-informed neural networks by Gradient-Guided Gaussian Adaptive Sampling (3GAS-PINNs)
Core Problem: Physics-informed neural networks (PINNs) provide a mesh-free framework for solving partial differential equations, yet their performance in nonlinear problems is often limited by slow convergence, gradient imbalance, and insufficient resolution to capture localized intermittent structures such as shock waves[1].
Key Innovation: Physics-informed neural networks (PINNs) provide a mesh-free framework for solving partial differential equations, yet their performance in nonlinear problems is often limited by slow convergence, gradient imbalance, and insufficient resolution to capture localized intermittent structures such as shock waves[1].
124. Applying foundation model embeddings towards urban livability evaluation
Core Problem: While accurate measurement of socioeconomic indicators remains challenging in data-scarce regions, which limits policy interventions and resource allocation, high-resolution geospatial data is widely available and can contain information on various livability statistics.
Key Innovation: We investigate which physical features are encoded within foundation model embeddings, such as AlphaEarth, AnySat, and TerraMind, and provide a systematic framework for identifying the most predictive geospatial indicators. Additionally, we demonstrate how to leverage foundation model embeddings to enhance prediction performance for these outcomes.
125. LogiScope-VQA: Benchmarking Vision-Language Models for Logistics Hazard Identification in Industrial Scenarios
Core Problem: However, the scarcity of real industrial data, tightly coupled to commercial terms, significantly hampers further advancement.
Key Innovation: However, the scarcity of real industrial data, tightly coupled to commercial terms, significantly hampers further advancement. We additionally reveal the pervasive security bias issue that impedes LLMs' practical deployment in real-world settings.
126. Development and Validation of a Physics-Guided Machine Learning Extrapolation Framework Using a Classical Transient Diffusion Benchmark
Core Problem: Machine learning models used in engineering are typically trained within limited operating ranges, yet reliable predictions are often required beyond these domains.
Key Innovation: To address this limitation, a novel extrapolation framework is integrated with established machine learning architectures to enable accurate and physically consistent predictions beyond the training domain. The results demonstrate accurate and physically consistent predictions beyond the training domain, highlighting the framework's potential for engineering applications where data availability is limited.
127. Structure-Aware Unsupervised Anomaly Detection for Spacecraft Telemetry with Adaptive EVT Thresholding
Core Problem: These requirements are rarely met in practice.
Key Innovation: We propose an unsupervised, deployment-ready framework that produces predictions from the second month of operation without any labels, prior fault knowledge, or mission-specific tuning. On the ESA Anomalies Dataset (ESA-AD), it achieves F0.5}=0.700 on Mission~1 and F0.5}=0.698 on Mission~2 under strict chronological evaluation.
128. Field-level prediction of mid-plane stress tensor fields in concrete target penetration: a cross-velocity graph neural operator surrogate
Core Problem: Although the impact resistance of concrete has been studied extensively, a framework linking mesoscale heterogeneity to full-field stress-tensor prediction has been lacking.
Key Innovation: Although the impact resistance of concrete has been studied extensively, a framework linking mesoscale heterogeneity to full-field stress-tensor prediction has been lacking.
129. Predicting Estimated Times of Restoration for Electrical Outages Using Longitudinal Tabular Transformers
Core Problem: Utilities publish Estimated Times of Restoration (ETRs) for customer-facing storm outages, and their accuracy governs whether customers can make sound decisions about food, medical equipment, and relocation.
Key Innovation: We reformulate ETR prediction as longitudinal tabular regression and introduce a Longitudinal Tabular Transformer (LTT), an axial-attention model that consumes the revisions preceding a prediction and issues a refined estimate at every one. Stratification by revision index shows that LTT error is largest at the first revision, where no history is available, and falls monotonically as revisions accumulate.
130. Temporal horizons in forecasting: a performance-learnability trade-off
Core Problem: However, we also prove that the loss landscape becomes rougher as the training horizon grows, making long-horizon training inherently challenging.
Key Innovation: In this work, we address this question by analyzing the relationship between the geometry of the loss landscape and the training time horizon. Our results provide a principled foundation for hyperparameter optimization in autoregressive forecasting models.
131. A Farewell to the Bias-Variance Tradeoff? An Overview of the Theory of Overparameterized Machine Learning
Core Problem: The last decade of progress in machine learning (ML), especially the deep learning era, has raised a number of scientific questions that challenge the longstanding dogma of the field.
Key Innovation: One of the most important riddles was the good empirical generalization of overparameterized models. Indeed, the discovery of the double descent phenomenon has revealed that highly overparameterized models can improve over the best underparameterized model in test performance.
132. Robustness of shallow graph embedding methods for community detection
Core Problem: However, the impact of perturbations on the performance of these methods remains relatively understudied.
Key Innovation: This study investigates the robustness of shallow graph embedding methods for community detection in the face of network perturbations, specifically node deletions. Through experiments conducted on both synthetic and real-world networks, the study reveals varying degrees of robustness within each family of shallow graph embedding methods.
133. A Jump-Diffusion Framework for Irregular Time Series Generation
Core Problem: We propose a framework for generative modeling of continuous-time processes from irregularly and asynchronously recorded data.
Key Innovation: We propose a framework for generative modeling of continuous-time processes from irregularly and asynchronously recorded data.
134. DFNN: A Deep Fréchet Neural Network Framework for Learning Metric-Space-Valued Responses
Core Problem: Regression with non-Euclidean responses-e.g., probability distributions, networks, symmetric positive-definite matrices, and compositions-has become increasingly important in modern applications.
Key Innovation: In this paper, we propose deep Fréchet neural networks (DFNNs), an end-to-end deep learning framework for predicting non-Euclidean responses-which are considered as random objects in a metric space-from Euclidean predictors. We further establish rigorous generalization guarantees for DFNNs and derive corresponding risk bounds, providing, to the best of our knowledge, the first such theoretical results for deep learning.
135. Study on the influence of arrangement patterns on wave attenuation characteristics of mangroves under regular waves based on SPH method
Core Problem: As an essential coastal ecological protection system, mangroves are capable of wave dissipation, shore protection and biodiversity conservation.
Key Innovation: To explore how spatial arrangements regulate the attenuation of regular waves by mangroves, this study employs the open-source code DualSPHysics. The results demonstrate that layout patterns exert a remarkable influence on wave attenuation performance under fixed plant density and bandwidth.
136. Numerical study on the hydrodynamic characteristics of a novel breakwater-reef assembly under wave and current actions
Core Problem: The integration of breakwaters with artificial reefs possesses combined advantages in protecting coastal lines and enhancing marine ecosystems.
Key Innovation: In light of this, the present study proposes a novel ecological wave-dissipating structure that combines a floating box-horizontal plate breakwater with a submerged artificial reef.
137. An ordinal threshold-guided framework for vessel accident risk prediction and high-risk vessel identification
Core Problem: However, severe class imbalance often causes conventional prediction methods to favor the majority low-risk class and underestimate vessels at high risk.
Key Innovation: Hence, this study proposes an ordinal threshold-guided framework combining CatBoost and SHAP (OTCSF) for vessel accident risk prediction and high-risk vessel identification. Results show that OTCSF achieves comparable overall performance while improving the recognition of vessels at medium or high risk and reducing severe underestimation and regulatory cost-sensitive loss.
138. Physics-data fusion-based dynamic model updating of semi-direct-drive offshore wind turbine gearboxes using full-scale test data
Core Problem: However, conventional models based on nominal parameters often fail to match measured responses because of uncertainties in mesh stiffness, support properties, damping and transmission errors.
Key Innovation: This study proposes a physics-data fusion-based dynamic model updating framework for semi-direct-drive offshore wind turbine planetary gearboxes using full-scale test data. Results show that the updated model significantly improves the agreement between simulated and measured vibration responses and more accurately captures the dynamic characteristics of the planetary gear train.
139. Investigation on dynamic response analysis of modular pontoon trestle subjected to multi-compartment flooding damage
Core Problem: To address the lack of clarity regarding mechanical property evolution and dynamic response mechanisms in modular pontoon trestles subjected to multi-compartment flooding damage, this study develops a mathematical model tailored for dynamic analysis of the loaded trestle following flooding in multiple compartments across multiple bridge segments.
Key Innovation: To address the lack of clarity regarding mechanical property evolution and dynamic response mechanisms in modular pontoon trestles subjected to multi-compartment flooding damage, this study develops a mathematical model tailored for dynamic analysis of the loaded trestle following flooding in multiple compartments across multiple bridge segments.
140. Numerical investigation on multi-compartment ship damage flooding with a scale effect correction method
Core Problem: Ship damage flooding poses a critical threat to maritime safety, yet the scale effects between model tests and full-scale ships remain a core bottleneck in damaged stability assessment and emergency response.
Key Innovation: This study employs a RANS-VOF framework with overset mesh, the realizable k-ε two-layer turbulence model, and the DFBI rigid-body motion model within STAR-CCM+. A multi-compartment flooding model based on the DTMB 5415 hull is then constructed with a realistic surface vessel layout, and full time-domain simulations are performed to reveal free-surface spreading, air pocket formation, and three-degree-of-freedom hull motion.
141. Integrating ozone-vegetation damage schemes into SSiB4/TRIFFID: evaluation of six parameterizations and refinement of ozone decay process across plant functional types
Core Problem: However, a significant spread remains when applying different schemes in various model frameworks.
Key Innovation: However, a significant spread remains when applying different schemes in various model frameworks. Our results show that O3 pollution led to approximately a 20% reduction in GPP during the 2010s, with discrepancies ranging from 15% to 31% across different schemes.
142. A strengthening yet less extreme Atlantic-European jet during winter
Core Problem: Nature Geoscience, Published online: 10 September 2026; doi:10.1038/s41561-026-02069-z Since 1725, weather extremes in Western Europe have been closely linked to the Atlantic-European jet stream, which is projected to strengthen by 2100 while becoming less extreme, according to daily sea-level pressure records and climate model simulations.
Key Innovation: Nature Geoscience, Published online: 10 September 2026; doi:10.1038/s41561-026-02069-z Since 1725, weather extremes in Western Europe have been closely linked to the Atlantic-European jet stream, which is projected to strengthen by 2100 while becoming less extreme, according to daily sea-level pressure records and climate model simulations.
143. Save the bogs, for peat’s sake
Core Problem: The verified record establishes the scope of save the bogs, for peat’s sake, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
144. Fencing threatens China’s protected areas
Core Problem: The verified record establishes the scope of fencing threatens China’s protected areas, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
145. Wars exacerbate the biodiversity crisis
Core Problem: The verified record establishes the scope of wars exacerbate the biodiversity crisis, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
146. China maps the deep scars of its geological past
Core Problem: The verified record establishes the scope of china maps the deep scars of its geological past, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
147. Data-driven approaches to rock typing in reservoir characterization: methodological advances, integration strategies, and challenges
Core Problem: This comprehensive literature review focuses on methodological advances in ML algorithms applied to rock typing, strategies for integrating these techniques with traditional petrophysical methods, and the emerging challenges that hinder widespread adoption.
Key Innovation: This is essential for accurate reservoir modelling, simulation, and enhanced hydrocarbon recovery.
148. Three decades of vegetation change in Jabel Marra: climate drivers and post-conflict vegetation dynamics
Core Problem: Understanding long-term vegetation dynamics in semi-arid and conflict-affected environments is important for sustainable land management and climate adaptation.
Key Innovation: This study investigated vegetation change and climate variability in the Jabel Marra region of Central Darfur, Sudan, during 1994-2024 using Landsat-derived Normalized Difference Vegetation Index (NDVI), CHIRPS precipitation, MERRA-2 temperature, and the Standardized Precipitation Evapotranspiration Index (SPEI). The annual NDVI showed a significant positive trend (Mann-Kendall Z = 4.78, p less than 0.001), with a Sen’s slope.
149. Changes in soil physico-chemical properties along elevation and precipitation gradients in Northeastern Syria
Core Problem: These factors generally affect soil formation and fertility, although data for northeastern Syria remain limited.
Key Innovation: This study assessed the physicochemical properties of three soil profiles (Al-Hasakah, Al-Qamishli, and Al-Malikiyah) in the Al-Hasakah Governorate (NE Syria) across elevation (300-598 m.a.s.l.) and precipitation gradients (280-650 mm). Al-Hasakah, the driest and lowest site, showed clay loam texture, the lowest clay content (30.4%), and low organic matter content (less than 0.77%).
150. Construction of composite biological soil crusts accelerates early-stage soil restoration in karst rocky desertification areas
Core Problem: Karst rocky desertification (KRD) represents one of the most severe forms of land degradation globally.
Key Innovation: In this study, we experimentally induced biological soil crusts (BSCs) through the use of moss, algae, and their co-inoculation, which represents a composite BSCs construction strategy, under both controlled and field conditions. Results showed that co-inoculation acted as the dominant driver of early-stage ecosystem restoration.
151. Spatiotemporal coupling between physical disaster impacts and social sentiment in urban flood events using large language models
Core Problem: Existing studies have largely examined physical disaster impacts or social sentiment separately, while their spatiotemporal coupling remains insufficiently quantified.
Key Innovation: This study proposes an LLM-driven spatiotemporal framework to investigate interactions between physical disaster impacts and social sentiment during urban flood events. The fine-tuned LLM achieved an F1-score of 0.895 for identifying posts containing both spatial information and physical disaster impacts, while sentiment scoring achieved an RMSE of 0.24 and an R² of 0.72.
152. TECHNOLOGICAL DIVERGENCE IN AMPHIBIOUS HOUSING: A COMPARATIVE ANALYSIS OF ACADEMIC FRAMEWORKS AND INDUSTRIAL PATENTS FOR FLOOD-RESILIENT HOUSING
Core Problem: Amphibious housing provides flood resilience by enabling structures to coexist with water rather than resisting it.
Key Innovation: This study investigates a technological divergence between academic research and industrial innovation that hinders the global scalability of these systems. Findings reveal a significant maturity gap: while academia provides the only field-validated deployments at TRL 7-8, industrial innovation is heavily concentrated at lower stages (TRL 2-3).
153. Multimodal learning for arcing detection in pantograph-catenary systems
Core Problem: However, electrical arcing at this interface poses serious risks, including accelerated wear of contact components, degraded system performance, and potential service disruptions.
Key Innovation: To address these challenges, we propose a novel multimodal framework that combines high-resolution image data with force measurements to more accurately and robustly detect arcing events. Through extensive experiments and ablation studies, we demonstrate that our framework significantly outperforms baseline approaches, exhibiting enhanced sensitivity to real arcing events even under domain shifts and limited availability of.
154. Predictive maintenance for aero-engine fleets integrating uncertainty-aware remaining useful life estimation and maintenance decisions
Core Problem: However, existing predictive maintenance studies generally focus on individual engines or treat remaining useful life (RUL) estimation and maintenance scheduling as separate processes, limiting their applicability to fleet-level operations under uncertain degradation conditions.
Key Innovation: To address this limitation, this study develops an end-to-end predictive maintenance framework that couples uncertainty-aware RUL estimation with multi-objective maintenance optimization for aero-engine fleets. Experimental results based on aero-engine datasets demonstrate that the proposed framework can effectively propagate RUL uncertainty into fleet maintenance decisions and generate high-quality maintenance plans for both.
155. Vegetation type overrides lithology in shaping seasonal water uptake patterns in a classical karst landscape
Core Problem: However, the relative importance of lithology and vegetation type in regulating seasonal plant water uptake remains poorly understood.
Key Innovation: However, the relative importance of lithology and vegetation type in regulating seasonal plant water uptake remains poorly understood. MixSIAR revealed that vegetation type exerted a stronger influence on seasonal plant water uptake than lithology.
156. Water-retention behaviour of waste tyre-CDG mixtures under wide suction range: Experimental Testing and Modelling Approach
Core Problem: However, the hydraulic response mechanism of such mixtures under unsaturated conditions remains unclear.
Key Innovation: This study investigated the compaction and water retention characteristics of CDG-rubber mixtures with rubber contents ranging from 0% to 30%, and established an adjusted soil-water characteristic curve (SWCC) model based on rubber content. The results indicate that increasing rubber content reduces the maximum dry density and increases the optimum water content, primarily attributed to the reduction in specific gravity of.
157. Long-term deformation of granite residual soil under cyclic loading in plane strain condition
Core Problem: Accurately predicting their long-term deformation remains challenging due to the complex influence of stress state and principal stress variations.
Key Innovation: This study presents a systematic experimental and modeling investigation into the cyclic behavior of GRS. Results show that the intermediate principal stress under plane strain enhances the vertical stiffness of GRS, andsuppresses the development of permanent strain.
158. Experimental investigation on sudden failure of segmental lining via scaled model tests
Core Problem: This manuscript investigates the failure mechanisms and root causes of sudden failure in shield tunnels through scaled model tests to understand why this failure occurs and how it can be prevented.
Key Innovation: The research findings indicate that: (1) The reserve safety factor for the ultimate limit state of segmental lining systems may be less than 1; that is, even if the segments and joints do not reach their yield strength, the lining system may lose bearing capacity due to a loss of flexural stiffness in the segment joints. (2) In the sudden failure mode of shield tunnels, the failure process of the full ring is dominated by the.
159. Vision-language models know more about agriculture than they show and rubric-grounded verifications close the gap
Core Problem: Vision-language models (VLMs) show promise for agricultural classification, but zero-shot performance on disease, pest, damage, quality, and species identification remains poor, and it is unclear whether this reflects weak visual features or a failure to connect them to domain knowledge.
Key Innovation: We build a benchmark of 116 datasets, 834 classes, and 8,324 images spanning these tasks to isolate where the gap arises. Linear probing shows VLM vision encoders already encode agricultural features nearly as separable as a self-supervised DINOv3 baseline, ruling out weak visual representations as the primary bottleneck.
160. StreetDiff: Multi-view Street Scenes Generation via Cross-view Consistent Multi-view Stable Diffusion with Structure Prompts
Core Problem: Multi-view diffusion models have shown strong performance in scenes with strong geometric priors and sparse semantics, such as indoor rooms or simple outdoor environments (e.g., fields, courtyards).
Key Innovation: To address this limitation, we propose StreetDiff, a multi-view diffusion framework that explicitly enforces cross-view alignment during denoising. By injecting structured alignment constraints without modifying the diffusion backbone, our framework achieves robust cross-view coherence in challenging urban street scene generation tasks.
161. Regularized Estimation and Feature Selection in Mixtures of Generalized Linear Experts
Core Problem: Sparsity is induced in both the gating network and the experts through ℓ1 penalties, and the penalized log-likelihood is maximized by a proximal Newton-EM algorithm whose M-step reduces to weighted Lasso problems with closed-form coordinate-ascent updates.
Key Innovation: We propose a regularized maximum likelihood framework for simultaneous parameter estimation and feature selection in MoE whose experts belong to the generalized linear model family, covering Gaussian, Poisson and multinomial responses within a single formulation.
162. Numerical study on vortex-induced vibration and aerodynamic sound radiation of an elastically mounted cylinder of mass ratio m*=2.4
Core Problem: The vortex-induced vibration (VIV) of an elastically supported circular cylinder (mass ratio m* = 2.4) subjected to a uniform incoming flow is numerically investigated using the Stress-Blended Eddy Simulation (SBES) over the reduced velocity range Ur = 1.33-14 (Re D = 1.37×104-1.44 × 105).
Key Innovation: The maximum vibration amplitude A*max = 0.962 in the upper branch agrees well with experimental measurements, while the lower-to-desynchronization transition amplitude is slightly overpredicted due to the inherent limitations of the 2D framework.
163. Aerodynamic forces and power extraction of flapping-foil energy harvesters under discrete gust inflow
Core Problem: However, most previous studies have considered steady or harmonically varying inflow conditions and have therefore not addressed the response to finite-duration discrete gust disturbances.
Key Innovation: In this study, a controlled numerical framework based on a 1 − c o s i n e discrete gust model is developed to evaluate FFEH performance. The results show that gust factor primarily governs the magnitude of aerodynamic excitation, whereas gust duration influences the persistence of the aerodynamic response.
164. Suppression of tip leakage vortex and flow-induced noise in a pump-jet propulsor via circumferentially distributed axial duct grooves
Core Problem: Systematic multi-objective optimization balancing hydrodynamic and acoustic performance is also lacking.
Key Innovation: This study introduces circumferentially distributed axial grooves on the inner duct wall to perturb tip leakage vortex (TLV) evolution under non-cavitating design conditions. Results indicate that the grooves fragment the TLV and attenuate its unsteady TKE, consequently reducing mid-to-high frequency broadband radiated noise.
165. Energy-efficient path planning for UUVs in complex ocean currents via flow-aware graph attention pointer networks
Core Problem: To address this challenge, we propose a Flow-Aware Graph Attention Pointer Network (FA-GATPN).
Key Innovation: To address this challenge, we propose a Flow-Aware Graph Attention Pointer Network (FA-GATPN). The results show that FA-GATPN achieves lower energy consumption than A*, RRT*, GA, ACO, CNN, and GCN baselines.
166. Geometry-guided and interaction-aware multi-agent reinforcement learning for cooperative multi-USV encirclement
Core Problem: However, cooperative encirclement involving multiple USVs remains challenging in real-world marine environments due to island-reef obstacles and dynamically maneuvering targets.
Key Innovation: To address this issue, we propose GHA-MAPPO, a geometry-guided and interaction-aware framework integrating a graph neural network (GNN) with multi-agent proximal policy optimization (MAPPO).
167. Environment-aware deep ensemble-guided ant colony optimization for multi-UUV path planning
Core Problem: However, most metaheuristic approaches rarely incorporate explicit environmental awareness.
Key Innovation: Therefore, we propose the Tri-Layer environment-aware Deep Ensemble Network (Tri-EDEN), a framework in which an environment-aware deep ensemble guides the Enhanced Ant Colony Optimization (EACO), and the three layers jointly solve the multi-UUV path planning problem. Comprehensive experiments involving 20 scenarios show that Tri-EDEN attains the lowest travel time in all cases and achieves a relative travel-time reduction of.
168. Lightweight optimization design of stiffened plate structures based on the plastic failure mechanism
Core Problem: Naval architecture and marine engineering must resolve two pivotal challenges: lightweight design and the accurate evaluation of the ultimate load-bearing capacity of stiffened plates under lateral loading.
Key Innovation: This study elucidates the global plastic failure mechanism of stiffened plates under lateral loading through a theoretical analysis and nonlinear finite element simulations. The theory-based optimization achieves moderate weight reduction while offering high computational efficiency and conservative outcomes.
169. Simultaneous tuning of collective and individual pitch controllers for blade load reduction and power regulation in offshore wind turbines
Core Problem: However, practical implementations exhibit residual coupling effects.
Key Innovation: This study proposes a multi-objective optimisation framework for the simultaneous tuning of CPC and IPC parameters to assess the trade-off between blade fatigue reduction and power regulation performance. Results show that simultaneous tuning of conventional CPC-IPC expands the achievable performance space, achieving average improvements of approximately 7-8% in blade fatigue loads and 23-24% in power regulation compared to.
170. RIST-Net: A risk-aware spatiotemporal interaction and risk-guided multi-scale trend decoding network for vessel trajectory prediction
Core Problem: However, in complex multi-vessel traffic scenarios, the future motion of a target vessel is affected not only by its own historical states but also by the interaction risks posed by neighboring vessels and the temporal evolution of such risks.
Key Innovation: Meanwhile, short-term motion continuity, medium-term interaction responses, and long-term trend stability in long-horizon prediction have not yet been sufficiently modeled within a unified framework. Experimental results obtained from five repeated runs on three real-world AIS datasets show that RIST-Net achieves the best performance in terms of MAE, RMSE, ADE, FDE, and FD.
171. Daily briefing: OpenAI claims a huge maths breakthrough
Core Problem: Nature, Published online: 09 September 2026; doi:10.1038/d41586-026-02867-w The AI giant claims to have solved one of maths’ trickiest problems, but questions swirl over should get the credit.
Key Innovation: Plus, an atlas of every single DNA letter mutation in humans and why more bioengineered microorganisms aren’t making it out of the lab.
172. Letting go of certainty
Core Problem: The verified record establishes the scope of letting go of certainty, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
173. Catching a glimpse of complex relaxation
Core Problem: The verified record establishes the scope of catching a glimpse of complex relaxation, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
174. NSF looks for technology fixes to ease water crisis along Colorado River
Core Problem: The verified record establishes the scope of nSF looks for technology fixes to ease water crisis along Colorado River, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
175. Scientists question U.S. Forest Service’s rationale for rescinding ‘roadless rule’
Core Problem: The verified record establishes the scope of scientists question U.S. Forest Service’s rationale for rescinding ‘roadless rule’, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
176. An agenda for implementation science in education
Core Problem: The verified record establishes the scope of an agenda for implementation science in education, but the accessible metadata do not expose the study motivation or boundary conditions.
Key Innovation: The title, authors, source and publication identity are verified; methods and quantitative outcomes are not asserted because a reliable abstract was unavailable.
177. Dynamic risk propagation and resilience in the global semiconductor trade network: A dual-modulated SIR model
Core Problem: The semiconductor industry underpins modern digital, energy, and defense systems, yet its global supply chain remains geographically concentrated and structurally rigid.
Key Innovation: We therefore develop a three-stage diagnostic framework using UN Comtrade data for 2018-2024 across silicon wafers, photographic chemicals, production equipment, and integrated circuits. Joint importance-resilience classification then reveals chokepoints invisible to either dimension alone.
178. Balancing Risks of Mission Failure and System Loss in Redundant System with Overlapping Rescue Procedures and Common Rescue Resource
Core Problem: Redundant systems are widely employed to mitigate the risk of mission failure.
Key Innovation: A probabilistic framework is developed to derive the mission success probability, expected cost of component losses, and normalized expected damage (NED), explicitly capturing interactions among mission continuation, RP execution, and resource contention.