TerraMosaic Daily Digest: August 6, 2026
Daily Summary
Slope studies now connect observable state variables to failure mechanics across field, regional and forecasting scales. At a vegetated shallow landslide, water content, suction and electrical conductivity were coupled to root-modified strength and local safety factors, making electrical observations mechanically interpretable rather than merely correlative. Compression tests on discontinuously cracked rock bridges add a failure-scale view: acoustic emission indicates shear-dominated rupture, while digital image correlation resolves a near-circular displacement field immediately before failure, consistent with circular slip-surface development in locked landslides. Spatial models add complementary context: a 3D convolutional network preserves neighbouring terrain conditions, a SHALSTAB mask constrains five-class susceptibility labels, and InSAR velocity improves urban susceptibility mapping by identifying actively deforming slopes. A new northern Noto dataset links legacy and 2024 event inventories to 3 m PlanetScope imagery, while a cracked tropical highway slope remains near failure across rainfall scenarios and develops shallow and approximately 8 m slip surfaces, with a local warning threshold near 300 mm over five days.
Volcanic and seismic work converts long records and process-scale observations into testable forecasts. A prospective California experiment evaluated 51,625 next-day forecasts from 27 earthquake models against 597 events; most reproduced event counts and magnitudes, but only selected operational-type models remained consistently compatible with the observations. At Stromboli, deformation during ordinary explosions, paroxysms and lava effusion is traced to a common pressurized reservoir about 280 m beneath the vent. At Mutnovsky, shallow 1.5–5 Hz tremor began about ten days before hydrothermal explosions. Bubble-scale viscoelastic failure predicts critical decompression rates for magma fragmentation, and crystal diffusion at Aira Caldera limits eruptible rhyolite storage to submillennial intervals, supporting brief rejuvenation from predominantly cold storage.
Hydroclimatic work shows that flood hazard cannot be inferred from event frequency alone because the boundaries that convert water into inundation are also changing. A global factual-counterfactual reconstruction attributes an increase in tropical-cyclone surge-flood extent from about 5% in 1950-1954 to about 18% in 2021-2025 to historical sea-level rise. Across 479 Brazilian rivers, changes in channel conveyance exert a larger influence on flood hazard than changes in flow frequency at most stations, chiefly through shifts in bed elevation. Kazakhstan's 2024 floods provide a complementary cold-region mechanism: snowmelt peaked before soil thaw, when ice-filled pores restricted infiltration. Forecasting studies then address decision scales directly, generating climate-conditioned flood sequences across more than 100 Mississippi Basin sites and showing that street-scale transfer learning is most useful before a new sensor has recorded roughly 18 flood events.
Key Trends
Five connected shifts link observable slope state, spatially informed susceptibility, prospective forecast tests, process-based volcanic interpretation and explicit treatment of changing flood boundaries.
- State Observations Are Being Translated into Failure Metrics: Electrical conductivity is mapped to suction, strength and local safety factor; digital image correlation resolves pre-failure deformation around locked rock bridges; InSAR velocity is introduced directly into susceptibility; and physics-constrained decomposition separates reservoir-driven displacement from trend and noise.
- Spatial Context and Physical Priors Are Reshaping Susceptibility Mapping: Neighbourhood patches, SHALSTAB-constrained class labels, deformation fields and attention-weighted multisensor ensembles replace pointwise binary classification with representations that retain terrain structure, active motion and failure-state gradients.
- Forecast Skill Is Being Tested Prospectively and Under Data Scarcity: California earthquake models are judged against a decade of unseen observations, flood-transfer experiments identify the sample size at which reuse loses its advantage, and storm-surge emulators are compared specifically at rare peaks rather than only at average conditions.
- Volcanic Signals Are Being Tied to Mechanism: Ground deformation, shallow tremor, bubble-scale melt failure and crystal diffusion constrain reservoir pressure, hydrothermal activation, fragmentation thresholds and the duration of eruptible magma states.
- Changing Boundaries Are Entering Flood Attribution: Counterfactual sea-level simulations, multidecadal channel-geometry records and coupled snowmelt-soil diagnostics separate changes in forcing from changes in the coastal, fluvial and frozen-soil states that convert forcing into flood extent.
Selected Papers
The leading papers connect direct measurements and counterfactual tests to hazard mechanism: a rooted shallow-slope study translates electrical and hydraulic observations into strength and local safety factors, rock-bridge experiments resolve shear rupture and incipient circular-slip deformation, a decade-long prospective experiment tests operational earthquake forecasts against unseen events, and a global reconstruction isolates the coastal area flooded because historical sea-level rise had already occurred. Four volcanic studies constrain reservoir pressure, hydrothermal precursors, magma fragmentation and the lifetime of eruptible rhyolite, while regional landslide mapping and extreme-rainfall field evidence extend process interpretation from individual sites to inventories, susceptibility zones and cascading sediment transfer.
1. Assessment of rooted slope instability using field hydraulic and electrical data
Core Problem: Electrical measurements can track water redistribution in soil, but root-modified hydrology and strength make them difficult to translate into the stability of vegetated shallow slopes.
Key Innovation: Field water-content, suction and conductivity records are embedded in root-aware hydro-mechanical relations and COMSOL simulations; modelled conductivity, displacement and local safety-factor trends agree with observations and conductivity-derived shear strength.
2. Select earthquake forecasting models demonstrate consistency with prospective decadal observations in California
Core Problem: Operational earthquake forecasts require prospective tests against observations that were unavailable when the models were developed, rather than retrospective fit alone.
Key Innovation: Twenty-seven models generated 51,625 next-day California forecasts over 2007–2018 and were tested against 597 earthquakes of Mw ≥ 3.95; the study identifies which operational-type formulations sustain consistency and releases common benchmarks and software.
3. Pressurization of Shallow Magma Reservoir Preceding Basaltic Eruptions at an Open‐Vent Volcano
Core Problem: Ordinary explosions, paroxysms and lava effusion at open-vent volcanoes differ greatly in intensity, leaving uncertain whether they engage distinct plumbing systems.
Key Innovation: A deformation-constrained conduit model attributes all three Stromboli eruption styles to pressure changes in the same gas-poor, crystal-rich reservoir about 280 m below the vent and links inflation or deflation to discharged magma volume.
4. Seismic Tremor as a Precursor to Hydrothermal Explosions Near Mutnovsky Volcano (Kamchatka)
Core Problem: Hydrothermal explosions can threaten geothermal infrastructure with little warning, and their preparatory seismic signal is poorly documented.
Key Innovation: Covariance analysis of 15 stations detects shallow 1.5–5 Hz tremor beneath Mutnovsky's northeastern slope beginning about ten days before the February 2024 explosions, consistent with hydrothermal activation by magmatic-fluid influx.
5. A Microphysical Model for Predicting Magma Fragmentation
Core Problem: Existing magma-fragmentation criteria omit bubble-scale melt physics and disagree with new decompression experiments.
Key Innovation: A viscoelastic failure model for melt surrounding rapidly expanding bubbles predicts whether experimental magma fragments and estimates critical decompression rates across magma types that agree with natural chemical and textural proxies.
6. Submillennial timescales of eruptible rhyolite magma: Converging evidence from Aira Caldera, Japan
Core Problem: Whether large rhyolitic reservoirs remain eruptible for long periods or are briefly remobilized from crystal-rich mush is unresolved.
Key Innovation: Concordant quartz and plagioclase diffusion times at Aira Caldera limit pre-eruptive residence to less than 1-2.5 kyr and the 400 km³ Aira-Tanzawa magma to less than 1 kyr, supporting short-lived eruptibility after rejuvenation.
7. Characteristics and formation mechanisms of catastrophic geological hazards triggered by extreme rainfall in the Yanshan Mountains, China, in July 2025
Core Problem: The July 2025 Yanshan disaster requires a mechanistic explanation of how exceptional rain was converted into clustered geological hazards across contrasting slopes and land uses.
Key Innovation: Field investigation of seven representative sites links prolonged extreme rainfall, carbonate-strata controls and shallow failures that supplied debris flows; it also shows that vegetation alone did not prevent failure and that terrace modification increased mobile sediment.
8. Increase of tropical cyclone-induced coastal flooding due to historical sea level rise
Core Problem: The global contribution of observed historical sea-level rise to tropical-cyclone surge inundation has not been isolated from storm variability.
Key Innovation: Shallow-water simulations of all IBTrACS landfalls from 1950-2025 compare factual sea levels with a no-trend counterfactual, attributing an increase in mean-tide flood extent from about 5% to 18% and showing strongly nonlinear expansion under 1 m of local rise.
9. Remote Sensing-Based Detection and Statistical Characterization of Large Landslides in the Transylvanian Basin
Core Problem: Large Quaternary landslides in the Transylvanian Basin lack a consistent regional inventory and morphometric characterization.
Key Innovation: Remote-sensing interpretation maps 171 landslides larger than 0.5 km², including two exceeding 10 km², and quantifies their geometry and spatial associations while preserving subdued morphology on apparently stable surfaces.
10. Attention‐Based Stochastic Simulation of Climate‐Informed Spatiotemporal Flood Risk for Multisite Insurance Portfolios
Core Problem: Flood portfolios need coherent interannual-to-decadal scenarios across sites, a horizon not covered by seasonal forecasts or long-range climate projections.
Key Innovation: Attention-based climate analogues and stochastic multivariate sequences generate linked flood frequency, intensity and duration at more than 100 Mississippi Basin sites, with explainable attribution and wavelets relating event clusters to large-scale climate modes.
11. Applying Transfer Learning for Street‐Scale Nuisance Flood Forecasting in Coastal‐Urban Environments
Core Problem: Street-scale flood models are difficult to deploy at newly instrumented locations where local event records are short.
Key Innovation: Norfolk experiments show that full-weight transfer of an LSTM is most beneficial when the target sensor has fewer than about 18 flood events; output-layer-only adaptation is inadequate, and the advantage fades as local data accumulate.
12. Long-duration oscillation of a tsunami in the west Japan region following the 21 December 2010 Mw 7.4 Bonin Outer-rise earthquake
Core Problem: The 2010 Bonin tsunami produced damaging, long-lived harbour oscillations that standard propagation models do not fully explain.
Key Innovation: Observation-constrained simulations identify shelf-edge waves and bay resonance sustaining oscillations beyond nine hours and show that nonlinear land-sea boundaries and bottom-friction choices materially alter coastal waveforms.
13. Drivers of flash flood frequency and intensity in the United States: a quantitative analysis of hydrometeorological interactions
Core Problem: Continental flash-flood trends reflect interacting precipitation, temperature, evapotranspiration and urbanization, but their relative effects are difficult to separate.
Key Innovation: More than three decades of hourly observations from 294 catchments identify precipitation as the primary control, quantify suppressing effects from warming and evapotranspiration, and associate a 10% impervious-area increase with an approximately 20% higher peak response.
14. Storm surge dynamics in the northern Adriatic Sea: comparing AI emulators with high-resolution numerical simulations
Core Problem: Operational storm-surge emulation must reproduce rare peaks without the training and inference burden of increasingly elaborate neural architectures.
Key Innovation: Across Adriatic tests, multiple linear regression paired with the MADc2 loss matches advanced neural models at extreme water levels while running orders of magnitude faster, separating the value of the loss function from that of model complexity.
15. From Multisensor Fusion to Intelligent Geospatial Monitoring: Emerging Architectures for Geotechnical Hazard Assessment
Core Problem: Geotechnical monitoring remains fragmented across SAR, optical, LiDAR and environmental streams, limiting temporal reasoning and transfer between sites.
Key Innovation: This review organizes the transition from engineered sensor fusion to multimodal transformers, self-supervised representation learning and geospatial foundation models, while isolating unresolved constraints in uncertainty, explainability, compute and environmental generalization.
16. Introducing an automated avalanche debris detection framework from sentinel-2 imagery using multi-index spectral fusion and U-net optimisation
Core Problem: Avalanche debris is difficult to distinguish from surrounding snow in rugged, variably illuminated satellite scenes.
Key Innovation: An eight-channel Sentinel-2 U-Net combines spectral indices with stepwise optimization and independent-scene testing; the final model reaches a mean Dice of 0.681, and PlanetScope comparison yields object-level precision 0.61, recall 0.87 and F1 0.72.
17. Rainfall-induced instability of a cracked heterogeneous highway slope: Parametric finite element analysis with SPT-based residual strength
Core Problem: Rainfall thresholds based only on factor of safety can miss deep infiltration through cracks and uncertainty in heterogeneous tropical highway slopes.
Key Innovation: Finite-element, probabilistic and resistivity evidence resolves approximately 2 m and 8 m potential slip surfaces, keeps the slope near failure across scenarios and supports a site-specific warning threshold near 300 mm accumulated over five days.
18. A Statistical-Physical Prior-Guided Framework for Multi-Class Landslide Susceptibility Mapping
Core Problem: Binary susceptibility prediction followed by post hoc class division does not directly learn hazard gradients and can spread historical failures across overly broad high-risk zones.
Key Innovation: SHALSTAB-constrained five-class labels and random-forest learning place 91.8% of 1,407 historical landslides in a very-high class covering 13.5% of the area, raising AUC to 92.53% and improving concentration over binary and certainty-factor labels.
19. A Novel TabPFN-Based Framework for Landslide Susceptibility Assessment: Multi-Model Comparison in Qingchuan County, Sichuan, China
Core Problem: Landslide susceptibility models must capture nonlinear controls from limited tabular geoscience samples without losing interpretability.
Key Innovation: TabPFN is compared with four standard classifiers in Qingchuan County and attains the highest reported AUC, 0.7834; SHAP identifies elevation, lithology and distance to roads as the principal controls on its predictions.
20. ADASEL: an attention-based adaptive stacked ensemble learning framework for landslide prediction in complex geophysical systems
Core Problem: Real-time landslide sensors produce strongly imbalanced, site-dependent time series in which failures are rare and event timing uncertain.
Key Innovation: ADASEL stacks seven diverse learners beneath an attention-based meta-model and improves macro-F1, recall and PR-AUC with low false-alarm ratios in within-site, bidirectional cross-site and leave-one-site-out tests across Indian Himalayan monitoring systems.
21. Development and application of a physics-constrained adaptive decomposition model for displacement sequences of step-like landslide
Core Problem: Step-like reservoir landslide displacement mixes long-term trend, periodic forcing and irregular motion, and conventional decomposition can separate them without physical meaning.
Key Innovation: An adaptively optimized variational decomposition adds correlations with reservoir level, rainfall and elevation as physical constraints, reconstructs displacement with about 0.5 mm RMSE and uses PyLith to support seepage-pressure triggering during drawdown.
22. Dynamic Earth Observation for Landslide Susceptibility Mapping Using Machine Learning and InSAR-Derived Deformation
Core Problem: Static conditioning factors cannot distinguish actively deforming urban slopes from dormant terrain with similar environmental attributes.
Key Innovation: Sentinel-1 SBAS velocity is validated against EGMS and added to a MaxEnt model for Cluj-Napoca, improving AUC from 0.809 to 0.833 and increasing correspondence between mapped landslides and high-susceptibility zones.
23. Landslide susceptibility assessment in Huizhou, South China: a comparison between 3-dimensional convolutional neural network and traditional point-based models
Core Problem: Point-based susceptibility models discard neighbourhood structure among terrain and environmental factors.
Key Innovation: A 3D convolutional network encodes factors as multichannel spatial patches, reaches an AUC of 0.93 in Huizhou and places 81.2% of historical landslides in a very-high class occupying 9.78% of the study area.
24. Landslide Inventories and PlanetScope Imagery for the Northern Noto Peninsula, Japan
Core Problem: Event-based landslide mapping in northern Noto needs co-registered imagery and an older inventory for separating persistent spatial controls from the 2024 earthquake-rainfall cascade.
Key Innovation: The open dataset pairs a legacy NIED inventory and an October 2024 interpreted event inventory with 3 m PlanetScope RGB imagery and georeferencing files for direct temporal and spatial comparison.
25. The number of clasts as quantity governing the mobility of rock avalanches
Core Problem: Rock-avalanche mobility increases with volume and decreases with grain size, but a common physical quantity explaining both relationships has remained contested.
Key Innovation: Numerical simulations and field data identify clast number, approximated by flow volume divided by median-clast volume, as the shared control and relate greater mobility to energy partitioning away from agitation and rotation and toward downslope translation.
26. Single Fault Earthquake Cycle Complexity
Core Problem: Irregular earthquake recurrence can arise from physical fault properties or numerical artefacts, complicating interpretation of earthquake-cycle simulations.
Key Innovation: Three-dimensional rate-and-state simulations separate controls from seismogenic width, nucleation length, dip and frictional heterogeneity and show that meshing choices can create spurious periodicity even when nominal resolution is adequate.
27. Diagnostics of the Hydrothermal Desynchronization of Snowmelt and Cryogenic Sealing of Soils During the Formation of Extreme Floods in Kazakhstan
Core Problem: The exceptional 2024 Kazakhstan floods require a process explanation for how rapid snowmelt generated runoff while soils remained frozen.
Key Innovation: A coupled heat-water-phase model evaluated at 65 stations reproduces soil state and water balance, then isolates the lag between snowmelt and thaw as a period of cryogenic pore blockage and suppressed infiltration; it is explicitly interpreted as a runoff diagnostic rather than an inundation model.
28. EERI Virtual Earthquake Reconnaissance Team (VERT) Phase 1 Report on the 2026 Uto, Japan Earthquake, in EERI 2026 Uto, Japan Earthquake Reconnaissance
Core Problem: Rapid post-earthquake decisions need a structured impact picture before field teams can collect ground-verified observations.
Key Innovation: The EERI VERT Phase 1 dataset assembles public evidence within 48 hours of the 28 July 2026 M6.8 Uto earthquake, covering source characteristics, aftershocks, response, tsunami and landslides, and impacts across buildings, transport, utilities, health and housing while retaining its preliminary status.
29. How changes in river geometry impact flood hazard in Brazil
Core Problem: Flood projections commonly treat river geometry as fixed, obscuring whether hazard changes arise from flow frequency or evolving channel capacity.
Key Innovation: A 1930-2018 analysis of 479 Brazilian rivers separates both effects and finds that conveyance change dominates at most stations; bed-elevation shifts explain much of the signal and are strongest in arid or heavily regulated basins.
30. Twin indicators of coastal vulnerability: land subsidence and mangrove extent change in Dili, Timor-Leste
Core Problem: Land subsidence and mangrove change are rarely quantified together in data-poor coastal cities, leaving their combined contribution to relative sea-level exposure unresolved.
Key Innovation: Sentinel-1 InSAR and Global Mangrove Watch provide the first paired assessment for Dili: mean subsidence is 0.53 cm per year, mangroves disproportionately occupy subsiding terrain, and projected sea-level rise plus subsidence exceeds 0.5 m across much of the city by 2100 without implying an unsupported causal link between the two indicators.
31. Study on the macroscopic deformation characteristics and failure mechanisms of rock bridge specimens under compressive shear stress state
Core Problem: Locked rock bridges govern the initiation of some landslides and tectonic ruptures, but laboratory tests have not reproduced their compressive-shear deformation path well enough to explain how a continuous slip surface develops.
Key Innovation: Uniaxial tests on specimens with discontinuous cracks combine mechanical response, acoustic emission and digital image correlation; shear-dominated rupture and a near-circular vertical-displacement field immediately before failure provide experimental evidence for circular slip-surface formation in homogeneous three-stage locked landslides.
32. Fluvial Geomorphic Resilience in Alluvial Mountain Rivers to Large Floods
Core Problem: Intensification of hydroclimatic volatility increases the potential for large floods to reshape rivers and threaten infrastructure.
Key Innovation: The authors assessed resilience at two scales: (1) a detailed geomorphic and hydraulic analysis of the June 2022 flood on East Rosebud Creek, Montana, which was part of broader flooding in the Greater Yellowstone Ecosystem, and (2) a multiriver comparison of 13 flood‐impacted alluvial mountain rivers (East Rosebud Creek plus 12 additional sites across North America and Europe). Understanding patterns of resilience can improve flood adaptation strategies and inform targeted river management.
33. A sediment record of floods ca. 1820–1950 for the River Isla, eastern Scotland
Core Problem: A 2.4 m long highly stratified sediment stratigraphy was recovered from an abandoned meander in the lowland reach of the River Isla in Perthshire, in eastern Scotland, 200 m from the river, closely connected to it through having no flood embankment.
Key Innovation: 1820–1950 corresponds in part to that from a new documentary record of flood frequency derived from local and regional newspapers 1820–1960.
34. High‐Resolution Imaging of Lithosphere‐Asthenosphere Structure in Northeast China Using Eikonal Tomography With Physics‐Informed Deep Learning
Core Problem: The lithosphere-asthenosphere transition beneath Northeast China's intraplate volcanoes is insufficiently resolved to distinguish lithospheric structure from deeper mantle controls on volcanism.
Key Innovation: Physics-informed eikonal tomography produces a 3D S-wave model to 120 km depth, resolving a thick high-velocity lithosphere beneath the Songliao Basin, an undulating transition zone and contrasting low-velocity structures beneath volcanic centres.
35. Seismic behaviour of scour protection rock berms
Core Problem: Rock scour protection around offshore monopiles can settle during earthquake-induced liquefaction, but design guidance for berms on liquefiable seabeds remains poorly constrained.
Key Innovation: Saturated dynamic-centrifuge tests isolate the effects of rock size and density, foundation-soil density and shaking intensity, then express the observed response in nondimensional relations intended for berm design.
36. Periodic extreme rainfall in a warmer climate due to stronger convectively coupled waves
Core Problem: Tropical regions may experience periodic extreme precipitation and suffer from associated periodic deluges in a warmer climate.
Key Innovation: Here, we conduct global-scale atmospheric model simulations with varying complexity.
37. Towards decision-oriented sub-basin prioritization of flood vulnerability: a case of Chitral River Basin, Hindu Kush Himalaya using an integrated framework
Core Problem: Mountain environments are particularly vulnerable due to complex terrain and limited data availability, underscoring the need for robust sub-basin-scale prioritization to support flood mitigation and resource allocation.
Key Innovation: The study developed an integrated framework to evaluate flood vulnerability across 11 sub-basins of the Chitral River Basin in the Hindu Kush Himalaya, Pakistan. The findings indicate that sub-basins characterized by complex drainage structures, reduced forest and wetland cover, and higher population exposure, particularly SB1 and SB11, followed by SB2 and SB3, are the most flood-vulnerable.
38. Detection and Tracking of Medicanes Through DeMeTrA Self-Supervised Vision Transformer
Core Problem: Medicanes are mesoscale cyclones that develop over the Mediterranean Sea and display tropical-like cyclone characteristics, including a warm core, spiral cloud organization, and deep convection over warm sea surfaces.
Key Innovation: To advance this research direction, this work introduces the Deep Learning Medicane Tracking (DeMeTrA) Algorithm, an end-to-end deep learning framework for medicane detection and rotation-center localization from SEVIRI Rapid Scan Airmass RGB imagery. On event-based splits, cyclone detection reaches 91% balanced accuracy on a balanced validation set and 89% on an unbalanced test set representative of operational conditions.
39. Seasonally Optimized Interferometric Network Stacking-InSAR for Subsidence Detection of Mined-Out Zones in Alpine Terrain: A Case Study of the Songhu Iron Mine, Xinjiang, China
Core Problem: Seasonal snow cover causes low coherence, decorrelation and phase-unwrapping errors that obscure mining subsidence in alpine InSAR time series.
Key Innovation: Optically derived snow-cover area partitions 590 interferograms into 168 high-quality summer pairs; the resulting Stacking-InSAR network delineates a coherent subsidence funnel with a maximum line-of-sight rate of -132.31 mm per year.
40. Assessment of urban vulnerability to floods under current and future demographic and climate scenarios
Core Problem: Climate change is increasing the frequency and magnitude of extreme precipitation, intensifying river flood risk.
Key Innovation: The study evaluates flood vulnerability in São Paulo’s Aricanduva River watershed under current and future demographic and climate conditions. Results showed that a 100-year flood could affect 3.5% of the population and 3.3% of households in 2022.
41. Initiation sequence of competing instabilities during high-pressure CO2 injection in overpressured reservoirs: fault shear reactivation, caprock tensile fracture initiation, and delayed pressure-relief feedback from blind faults
Core Problem: High-pressure injection for geological CO₂ storage may rapidly disturb the equilibrium between porepressure and in-situ stress in naturally overpressured reservoirs, thereby inducing caprock tensile fracture initiation and fault shear response.
Key Innovation: The study investigates a conceptual scenario characterized by an ultralow-permeability tight matrix, natural overpressure, a blind fault, and short-term strong pressure loading. The results show that, within the examined parameter range, all triggered cases exhibit caprock tensile fracture initiation before fault plastic response and significant permeability-enhancement feedback, and no fault-leading regime is observed.
42. Effects of varying incident wave inclination and azimuthal angles on multi-dimensional ground response analyses at the Delaney Park Downhole Array site
Core Problem: most prior two-dimensional (2D) and 3D GRAs assume vertically propagating, horizontally polarized shear-wave input, neglecting scenarios where seismic waves approach at inclined angles and from different azimuths.
Key Innovation: The study evaluates the effects of varying inclination and azimuthal input angles in 2D and 3D GRAs at DPDA to assess whether non-vertical and directional wave incidence increase apparent damping and improve agreement with observed ETFs.
43. Effect of laponite treatment on liquefaction-induced tunnel uplift: Evaluation based on multiple intensity measures
Core Problem: Soil liquefaction poses a significant threat to shallow-buried tunnels, which are susceptible to liquefaction-induced uplift during strong earthquakes.
Key Innovation: The study employs extensive ground motion records with varying characteristics as input excitations to assess the liquefaction mitigation effectiveness of laponite and further discusses its mechanism in mitigating liquefaction-induced tunnel uplift.
44. Application of GIS techniques for landslide susceptibility mapping in the mountainous regions of Fetakgomo-Tubatse Municipality
Core Problem: The study compares AHP, fuzzy logic, and XGBoost landslide-susceptibility models for mountainous terrain and engineered slopes along the R37 and R555 corridors in South Africa.
Key Innovation: Fuzzy logic produced the highest ROC-AUC (0.769), ahead of AHP (0.757) and XGBoost (0.739), with high-susceptibility zones concentrated on steep, weathered, structurally controlled, and road-modified slopes.
45. Unveiling landslide susceptibility in the mediterranean context: a weighted overlay analysis of Kaous municipality, NE Algeria
Core Problem: Kaous municipality combines steep slopes, weak lithologies and high rainfall but lacks a validated spatial basis for prioritizing landslide-prone terrain.
Key Innovation: A GIS analytical-hierarchy weighted overlay combines seven conditioning factors and classifies 54.85% of the municipality as high to extremely high susceptibility, while explicitly identifying inventory validation and additional geotechnical factors as remaining needs.
46. Biogeophysical Characterization of a Landslide Zone Using Remote Sensing for Disaster Risk Management at San Vicente Volcano, El Salvador
Core Problem: A landslide on San Vicente Volcano lacked integrated measurements of its steep geometry, displaced volume, vegetation disturbance and depositional pattern for local monitoring and risk management.
Key Innovation: UAV photogrammetry, GIS and Sentinel-2 NDVI characterize a 9,570.50 m² feature on slopes above 70 degrees and estimate about 191,410 m³ of mobilized material while mapping morphological and vegetation change.
47. Google Earth Engine Scripts and Supporting Data for Random Forest-Based Landslide Susceptibility Mapping in Hanang District, Tanzania
Core Problem: It includes the landslide inventory, conditioning-factor workflow, modelling data, predictor stack, and multicollinearity diagnostics needed to reproduce the analysis.
Key Innovation: This repository provides Google Earth Engine scripts and supporting data for a Random Forest landslide-susceptibility model in Hanang District, Tanzania.
48. Distinguishing compound and cumulative hazards using machine learning and fuzzy logic in multi-hazard susceptibility mapping
Core Problem: Machine-learning susceptibility maps for floods, avalanches, rockfalls, and landslides along the Haraz Road corridor in northern Iran are integrated with fuzzy AND, OR, and GAMMA operators.
Key Innovation: Random forest led flood and rockfall models, support vector machines led avalanche and landslide models, and the AND operator gave the strongest multi-hazard validation (ROC-AUC 92.20%; F1 93.83%) while GAMMA delineated broader cumulative-susceptibility zones.
49. Instability Risk of Submarine Hydrate-Bearing Slopes Under Thermal Disturbances
Core Problem: Thermal disturbance can destabilize submarine hydrate-bearing sediments by reducing hydrate stability, promoting dissociation, weakening hydrate-derived cementation, and increasing excess pore pressure.
Key Innovation: The study develops a probabilistic framework coupling a one-dimensional thermal–hydrate evolution model with an infinite slope stability formulation. The minimum factor of safety, FSmin, decreases from approximately 6.2, crosses the warning threshold of 1.30, and first reaches the critical threshold of 1.00 at approximately 4.6 ka after substantial hydrate system degradation.
50. Long-Term Evaluation of Satellite Precipitation Products for Extreme Rainfall and Water-Related Hazard Assessment Along a Mountainous Corridor in Northern Vietnam
Core Problem: Accurate rainfall information is essential for water-related hazard assessment in mountainous regions, where complex terrain and sparse gauge networks limit monitoring reliability.
Key Innovation: The study evaluated long-term rainfall characteristics and the performance of three satellite precipitation products—CHIRPS, GPM IMERG, and GSMaP—along the National Highway 6 corridor in northern Vietnam. Data from 2025 were additionally examined as an extreme-rainfall case study, supplemented by visual gauge–satellite comparisons; however, these data were not used as an independent satellite-validation period.
51. Reducing false alarms in urban flood detection: an enhanced NDWI (ENDWI) with Hybrid Max Fusion on Sentinel-2 Data
Core Problem: Optical urban-flood mapping produces false positives where built surfaces are spectrally confused with water.
Key Innovation: ENDWI, zero-preserving split normalization and maximum fusion with AWEIsh reduce the false-alarm rate to 2.99% on a Sentinel-2 case validated by 1,262 WorldView-4 reference points, with 94.50% precision and 82.65% overall accuracy.
52. Seismic Risk of Critical Transport Systems: Sensitivity Analysis of the Campania Railway Network to Bradyseism's Features
Core Problem: The study examines the impact of epistemic uncertainties on the seismic risk of a critical transport system, the Campania railway network, during the ongoing bradyseismic unrest at Campi Flegrei, south of Italy.
Key Innovation: A scenario‐based risk formulation is adopted, integrating fragility functions for tunnels and masonry arch bridges with a ground‐motion model tailored explicitly to volcanic areas. Increases in magnitude or ground‐motion median, or softer soils, markedly expand the network to be inspected, whereas lower component vulnerability, such as bridges and tunnels, reduces it.
53. Spatial Methods for Identifying Undocumented Historical Earthquake Damage
Core Problem: Historical earthquake records are inherently incomplete: many sites that were likely damaged were never documented, leaving spatial gaps in the macroseismic record.
Key Innovation: The study evaluates whether intensity values at unreported sites can be estimated from the spatial relationships of surrounding reports, framing the task as a spatial data imputation problem. Model performance was assessed with 5-fold cross-validation under random and spatial-block designs, using Mean Squared Error (MSE) and success rate, defined as the percentage of predictions falling within ±0.5 and ±1.0 intensity units of observed values.
54. Vision-Based Autonomous Quadrupedal Robot for Rapid Post-Earthquake Crack-Based Building Damage Detection
Core Problem: Post-earthquake crack inspection is hazardous and slow when engineers must enter potentially unstable reinforced-concrete structures.
Key Innovation: A quadruped carrying an RGB-D camera and edge-deployed YOLOv8n is trained on 3,255 annotated crack images; laboratory tests report precision, recall and mAP@0.50 above 85%, establishing a deployable inspection prototype rather than field-validated damage grading.
55. A GIS-based framework for identifying isolated communities and population-level accessibility assessment under compound multi-hazard conditions
Core Problem: Compound hazards can fragment road networks and isolate communities, but hazard-specific connectivity studies rarely quantify population access to hospitals, airports and emergency resources.
Key Innovation: A GIS exposure-network framework combines disconnected-component detection, road centrality and population-weighted accessibility; in coastal Oregon, flood-inclusive scenarios cause the greatest fragmentation and near-field tsunami-plus-flood scenarios cause the largest accessibility losses.
56. Statewide Probabilistic Flood Hazard Maps for Idaho, USA: Ensemble Mean Water Depth and Flood Probability Datasets Derived from the Torrent Hydraulic Model
Core Problem: Statewide flood planning needs spatially complete depth and probability estimates that propagate rainfall duration, return-period and precipitation uncertainty rather than a single deterministic scenario.
Key Innovation: The Idaho dataset aggregates 100 Torrent hydraulic realizations for each of 48 rainfall configurations into inundation probability and ensemble-mean depth layers distributed as 30 m Cloud Optimized GeoTIFFs.
57. Parametric Study on the Topographic Amplification Effect of Tunnel-Slope Under the Oblique Incidence of SV Wave
Core Problem: A numerical tunnel-slope model quantifies topographic amplification under vertical and obliquely incident SV waves.
Key Innovation: Amplification increases with incident frequency, obliquity, distance from the crest, and decreasing tunnel radius; omitting oblique incidence can underestimate peak horizontal and spectral acceleration amplification by 15.6% and 19.4%, respectively.
58. Simultaneous effusive and explosive activity in moderate-magnitude rhyolite eruptions: H2O and hydrogen isotopic investigation of the 1250 CE eruptions in Lipari, Sicily
Core Problem: Microtextures, water contents, and hydrogen-isotope ratios from pyroclastic and effusive products of three circa-1250 CE rhyolite eruptions on Lipari follow a single degassing trend.
Key Innovation: The evidence supports repeated, simultaneous explosive and effusive activity from one magma source and implies that the three youngest centres were co-eruptive during an island-scale fissure event.
59. Assessing the stratospheric temperature response to volcanic sulfate injections by Mt. Pinatubo: insights from the Interactive Stratospheric Aerosol Model Intercomparison Project
Core Problem: The scale of these impacts for varying injection amounts and heights on stratospheric temperature anomalies is still highly uncertain.
Key Innovation: Here we analyse specially designed chemistry-climate model experiments following the Historical Eruptions SO2 Emission Assessment Protocol (HErSEA) under the Interactive Stratospheric Aerosol Model Intercomparison Project (ISA-MIP).
60. Integrating land use scenarios with pluvial flood modelling to evaluate spatial development strategies for climate adaptation in the Stuttgart region
Core Problem: Climate change and urbanisation are exacerbating climate risk, including pluvial flooding.
Key Innovation: The study assesses impact of urban development on pluvial flood exposure in the Stuttgart Region, Germany. We develop and simulate two spatially explicit urban growth scenarios based on current spatial plans: (1) a business-as-usual scenario reflecting current development practices, and (2) a climate-sensitive scenario characterised by more compact development, increased permeable green space, and the integration of blue-green infrastructure.
61. Enhancing resilience through recovery modeling in multi-hazard contexts
Core Problem: Natural hazards striking in succession impose compound disruptions on communities, extending recovery timelines and challenging resilience planning.
Key Innovation: Here, we develop a probabilistic framework that integrates hazard sequencing, time between hazard events, and relative intensities to quantify their combined effects on the recovery of building structures. Our results demonstrate that accounting for multi-hazard interactions leads to exacerbated restoration times, informing adaptive policy measures that support efficient resource allocation.
62. Autonomous prediction framework for the seismic risk of RC girder bridge clusters considering fatigue and mainshock-aftershock sequences
Core Problem: A probabilistic framework combines bridge fatigue with mainshock-aftershock hazard, ground-motion monitoring from the 2008 Wenchuan and 2024 Ishikawa earthquakes, and field observations from 386 reinforced-concrete continuous girder bridges.
Key Innovation: Random forest, multilayer perceptron, convolutional, and recurrent models use ten influential features to predict updated seismic-fragility levels.
63. ClimEmpower River Flooding Risk Assessment Methodology and Maps
Core Problem: Regional adaptation programmes need comparable, reproducible estimates of river-flood depth and direct damage across jurisdictions with different exposure and vulnerability data.
Key Innovation: ClimEmpower combines JRC/CEMS depth maps, LUISA land use, JRC depth-damage functions and DamageScanner for five European regions, releasing scripts, inputs and hazard and damage rasters for 10-, 100- and 500-year return periods.
64. Coastal erosion and shoreline protection in West Africa: drivers, limits of current responses, and pathways toward integrated and sustainable coastal management
Core Problem: A structured review of 64 publications from 2000-2025 examines coastal-erosion drivers and protection performance in West Africa and the Gulf of Guinea.
Key Innovation: Hard structures can stabilize local shorelines in the short term but often shift erosion downdrift, disrupt sediment budgets, damage ecosystems, and impose maintenance and livelihood costs; the review calls for long-term monitoring, regional sediment budgets, life-cycle assessment, and more carefully evaluated nature-based and hybrid responses.
65. D.2.1.4 Flood exposure maps for Ravenna and Kaštela test sites - Interreg Italy-Croatia 2021-2027
Core Problem: Coastal emergency planning in Ravenna and Kastela requires return-period flood extents to be translated into the land uses and territorial assets exposed within each scenario.
Key Innovation: The STRENGTH dataset intersects 2023 total-water-level flood zones with regional land-use layers and supplies sector-specific exposure maps for 1-25-, 50-100- and 150-500-year classes.
66. D2.4.2 Database of observed flood events - Interreg Italy-Croatia 2021-2027
Core Problem: Local coastal-risk analysis is limited when observed floods and their documented impacts are dispersed across incompatible storm catalogues and administrative records.
Key Innovation: The STRENGTH project harmonizes event databases for Ravenna and Kastela, adapting established European coastal-storm catalogues to preserve storm characteristics and observed impacts for emergency planning.
67. Effects of Urban Gray–Green Spatial Morphology on Surface Runoff: A Case Study of Typical Flood-Prone Blocks in Shenyang, China
Core Problem: The nonlinear and interacting effects of three-dimensional building form and two-dimensional green-space morphology on block-scale runoff remain difficult to translate into urban flood-mitigation design.
Key Innovation: Random forest and SHAP analysis across six flood-prone Shenyang blocks resolves threshold and saturation effects in building congestion, height variation, green-space scale, boundary complexity and fragmentation, with R-squared values of 0.723-0.825.
68. Estimation of Site Effects and Seismic Attenuation (Q) Using Generalized Inversion in Southwestern Saudi Arabia: Implications for Seismic Risk Assessment
Core Problem: Generalized inversion of earthquakes larger than magnitude 3 recorded at 16 stations separates source, path, and site contributions in southwestern Saudi Arabia.
Key Innovation: Horizontal-to-vertical spectral ratios and inverted predominant frequencies characterize local amplification and support site classification for regional seismic-risk assessment.
69. Flood Hazard Mapping and Evacuation Route Planning Using Geographic Information Systems (GIS) in Malang City, East Java, Indonesia
Core Problem: Flooding is one of the most frequent hydrometeorological disasters in Indonesia and continues to threaten urban areas due to rapid land-use changes, increasing impervious surfaces, and inadequate drainage systems.
Key Innovation: Malang City, East Java, has experienced rapid urban development that increases flood susceptibility in several districts.
70. GIS-Based Multi-Criteria Flood Hazard Mapping Using a Scoring and Weighting Approach in Trucuk and Bojonegoro Districts, East Java, Indonesia
Core Problem: Trucuk and Bojonegoro Districts in Bojonegoro Regency, East Java, are low-lying areas that experience recurring inundation whenever the Bengawan Solo River overflows during the rainy season.
Key Innovation: The study maps flood-prone zones across the two districts using a Geographic Information System (GIS) so that flood risk reduction efforts can be directed more precisely.
71. Historical Flood Events and Documentary Sources for Holstebro, Denmark, 1878–2025
Core Problem: Long-term flood-frequency and impact analysis for Holstebro lacks an event chronology that retains explicit links to the documentary evidence supporting each historical flood.
Key Innovation: The dataset records 64 Storaa River floods from 1878-2025 and connects them to 692 sources through 556 event-source-role assignments, with relational tables, a codebook and methodological documentation.
72. Intra-block fault strain accommodation reshapes continental deformation of the Bayan Har block in Tibet
Core Problem: High-resolution InSAR and GNSS observations resolve deformation within the central-eastern Bayan Har block.
Key Innovation: Heterogeneous slip rates, locking depths, and distributed strain support a semi-confined, non-rigid bookshelf-fault model rather than rigid-block kinematics, changing how strain transfer and seismic potential are interpreted.
73. Enhancing wind‐induced erosion resistance of Indian Thar Desert sand using enzyme‐induced carbonate precipitation
Core Problem: Soil reinforcement is a potential solution to mitigate wind‐induced sand erosion.
Key Innovation: The present study explores the effectiveness of an alternative biologically inspired method, enzyme‐induced carbonate precipitation (EICP), in enhancing the stability and strength of aeolian sand. The study demonstrated that the EICP method effectively mitigates erosion up to a wind speed of 30 m/s and achieves optimal soil strength, with a maximum UCS value of 756 kPa and 5% calcite formation.
74. Gamma Rays Signal the Vulnerability of Forest to Extreme Drought
Core Problem: Extreme drought and heat are driving forest die‐off globally, yet regional‐scale mapping of subsurface water storage capacity, governing forest vulnerability, remains a major challenge.
Key Innovation: While GRS measures the upper ∼30 cm of soil, plot‐scale electrical resistivity tomography suggests that elevated 40 K signals shallow granite or gneiss bedrock, to depths reaching several tens of meters due to leaching of mobile elements, remnant bedrock minerals, erosional thinning of regolith and biological nutrient recycling from K‐rich saprolite.
75. Satellite Mapping of the Moulin Distribution Across the Entire Greenland Ice Sheet
Core Problem: The ice-sheet-wide distribution of moulins, which route surface meltwater to the bed and alter subglacial hydrology, has not previously been mapped consistently.
Key Innovation: Manual interpretation of 10 m Sentinel-2 imagery identifies 5,869 major moulins across Greenland; 88.5% are river-fed, density peaks near the margin, and 6.6% extend into the accumulation zone.
76. Automated Adaptive Baseflow Separation Across Australia's Catchments Using AutoVL
Core Problem: widely used digital filters such as the Lyne–Hollick (LH) algorithm, when applied with a fixed, uncalibrated parameter, can substantially underestimate baseflow and obscure non‐stationary streamflow responses.
Key Innovation: AutoVL reconstructs streamflow components using a variable‐leak signal formulation coupled with a statistical recession model that iteratively estimates governing parameters from the discharge record.
77. Beyond GRACE: Evaluating the benefits of NGGM and MAGIC for precipitation estimation over Europe
Core Problem: their coarse spatio-temporal resolution limits the applicability in many hydrological studies.
Key Innovation: The study employed the well-known SM2RAIN algorithm to estimate the precipitation accumulated between two consecutive TWS measurements. The results showed that SM2RAIN exhibited satisfactory performance at a daily temporal resolution, with a mean correlation coefficient of 0.86.
78. Global distribution and amplitudes of internal solitary waves revealed by SWOT
Core Problem: Yet, their global distributions and vertical imprints remain poorly characterized.
Key Innovation: Here, we develop an artificial intelligence framework to automatically detect ISWs worldwide.
79. Raman spectroscopy characterization and maturity evolution of organic carbon in shale-related fault rocks
Core Problem: Frictional heating from fault slip in organic-rich rocks can drive the structural reorganization of organic carbon (OC), revealing a tectonic control on the enhancement of thermal maturity.
Key Innovation: Here we combine Raman spectroscopy of fault rocks from distinct structural domains within the Lower Silurian Longmaxi Formation in Sichuan Basin with Raman maturity ( Rmc Ro ) calculations to constrain the impact of fault activity on organic carbon maturity (OCM).
80. Variability in geomorphic response to millennial-scale hydroclimatic shifts and fire-induced disturbance: a multi-proxy analysis of a peatland archive from South Africa's Mediterranean biodiversity hotspot
Core Problem: How millennial hydroclimatic shifts and fire disturbances alter sediment connectivity in South Africa's Mediterranean-climate landscapes remains poorly constrained.
Key Innovation: A 2,858-year peat archive combines grain size, pollen, charcoal and dung spores, showing that fire can precede sediment pulses but hydroclimatic forcing governs transfer and that organic sedimentation can recover after compounded disturbance.
81. SWENet: A multi-feature fusion network for natural and flooded water body extraction from high-resolution remote sensing imagery
Core Problem: High-resolution extraction of natural and flood-inundated water bodies often loses narrow channels, fragmented patches and boundary detail in spectrally complex scenes.
Key Innovation: SWENet fuses edge, spectral and texture cues through a transformer branch, bidirectional gated fusion and multiscale context, improving delineation of small fragmented and flood-affected water bodies across three evaluation datasets.
82. Data-driven insights into transitional slip behaviors of rock fractures
Core Problem: Although numerous theoretical and experimental studies have advanced our understanding of transitional slip behaviors, the underlying mechanisms remain insufficiently constrained to predict the onset of fracture instability and the associated induced hazards.
Key Innovation: Here we integrate laboratory friction experiments with a dual-stage attention-based recurrent neural network to investigate the dominant factors of transitional slip behaviors. The machine learning model is trained based on a set of observable factors, including slip velocity, acoustic emission (AE) energy, and time-since-previous-event, to predict time-to-failure while quantifying the relative importance of each factor during the slip process.
83. Modeling in-situ gas flow and coupled hydro-mechanical processes in the Opalinus Clay
Core Problem: Gas generated in low-permeability clay repositories can build pressure above the minimum principal stress, yet the conditions separating two-phase flow, dilatancy and fracture transport remain poorly constrained.
Key Innovation: TOUGH-FLAC couples multiphase flow and geomechanics against the Mont Terri gas-injection experiment, reproducing pressure evolution and testing permeability formulations that capture the onset of dilatant pathways.
84. Field experiments on vertical strain monitoring of PHC pipe piles using quasi-distributed fiber optic sensors
Core Problem: Vertical strain and load transfer during static pile jacking must be observed continuously because excessive strain can compromise prestressed concrete pile integrity.
Key Innovation: Six pairs of embedded fibre Bragg grating sensors monitored two full-scale 12 m piles and resolved axial force, shaft resistance and end resistance; the end-resistance ratio rose by up to 40.4% on entering stiff silt while measured strain stayed below the buckling threshold.
85. Evaluation of stereo-microphone impact echo signal features for machine learning-based detection of voids beneath concrete pavement
Core Problem: Subsurface void detection in pavement infrastructure using conventional single-sensor impact-echo testing exhibits fundamental limitations in spatial correlation analysis and automated classification.
Key Innovation: To address these limitations, this study employs stereo-microphone configurations integrated with supervised machine learning to establish quantitative relationships between void geometry and classification performance. The results reveal geometry-dependent transitions in classification performance, primarily governed by edge-reflection mechanisms.
86. Effects of soil structure on the cyclic and post-cyclic behaviour of a saturated clay under various stress states
Core Problem: structure effects on cyclic and post-cyclic behaviour under initially anisotropic stress states remain unexplored.
Key Innovation: The study examined the cyclic and post-cyclic responses of saturated compacted and intact completely decomposed metasiltstone (CDM) under various initial mean effective stresses, initial deviatoric stresses, and cyclic deviatoric stresses. The results show that intact specimens develop lower excess pore water pressure (PWP) and smaller cyclic axial strain than compacted specimens under identical stress conditions and comparable initial void ratios.
87. Quantitative in situ CT analysis of fatigue damage evolution in coal pillars under progressive and cyclic loading
Core Problem: the quantitative link between macroscopic fatigue behavior and internal fracture evolution in mining-disturbed coal remains unclear.
Key Innovation: Underground energy storage in coal mines requires rigorous assessment of the long-term structural stability of residual coal pillars. The results show that damage evolution varies with loading conditions.
88. Carbon capture biocementation pathways for sustainable geotechnical engineering: a review
Core Problem: Reported systems raise small-strain stiffness, capture carbon dioxide, and support applications from erosion control to liquefaction mitigation, while ammonia, treatment uniformity, scale-up, and life-cycle cost remain deployment constraints.
Key Innovation: This review compares MICP, EICP, carbonic-anhydrase biomineralization, denitrification, methane oxidation, and photosynthesis as carbon-capturing soil-improvement pathways.
89. Examination of hydro climatic trend and drought analysis for climate resilience in the eastern escarpment of Abaya Chamo sub-basin, Rift Valley Lake Basin, Ethiopia
Core Problem: A robust assessment of hydroclimatic variability is critical for strengthening drought resilience and optimizing water resource management in climate-sensitive and data-limited regions.
Key Innovation: The study provides one of the first comprehensive integrated hydroclimatic analyses of the eastern escarpment of the Abaya-Chamo sub-basin, evaluating multi-decadal trends in rainfall, temperature, streamflow, and drought through the Mann-Kendall trend test, Indicators of Hydrologic Alteration (IHA), and drought indices derived using DrinC software.
90. Evolution Characteristics of Extreme Precipitation in Northern China: A Case Study of Chaohu Lake Basin
Core Problem: Long-term changes in extreme precipitation around Chaohu Lake require joint evidence of trend, regime shift and exceedance probability rather than a single diagnostic.
Key Innovation: A trend-mutation-probability framework applied to 15 stations from 1961-2019 identifies a 1979 shift, a 2.4 mm per year basin-wide increase and a 154.6 mm rise in the 90th-percentile total after the transition.
91. Behavioral Coordination in Community Pluvial Flood Resilience: A Tripartite Evolutionary Game Analysis
Core Problem: Urban pluvial flood resilience relies not only on engineering defenses but also on effective behavioral coordination among local governments, property management companies (PMCs), and residents.
Key Innovation: The study addresses this dilemma by integrating prospect theory into a tripartite evolutionary game model. Simulation results show that actors initially remain inactive under uncertainty.
92. Remote Sensing-Based Assessment of Hydro-Climatic Variability and Spatial Drought–Extreme Rainfall Conditions in the Transboundary Tano River Basin, West Africa
Core Problem: Drought and rainfall-potential rankings remain stable under equal weights and 20% perturbations, and monthly CHIRPS agrees strongly with GPM IMERG (R-squared 0.881).
Key Innovation: CHIRPS rainfall and Landsat observations from 2000 to 2024 map rainfall, extreme oneand five-day totals, dry spells, vegetation, surface temperature, drought, and extreme-rainfall potential in the transboundary Tano basin.
93. Unveiling the indirect flood-control benefits of the three gorges project: a water conservancy general equilibrium analysis
Core Problem: Direct damage accounting misses flood losses propagated through supply chains and therefore understates the economy-wide protection supplied by large hydraulic projects.
Key Innovation: A water-conservancy general-equilibrium model estimates risk-dependent indirect benefits of the Three Gorges Project; under the high-risk case, avoided indirect GDP losses reach 323.667 billion yuan for the 2016 economy and 483.039 billion yuan for 2022.
94. Improving Accuracy of Low-Altitude Three-Dimensional Wind Field Reconstruction via Multi-Task Learning
Core Problem: The demand for high-precision 3D wind fields in low-altitude weather services is increasing.
Key Innovation: To improve reconstruction accuracy, the high precision of 2D near-surface single-level products and the vertical structure of 3D numerical model outputs are combined to establish a mapping between 2D observations and 3D atmospheric fields, offering a new approach for 3D wind field reconstruction.
95. Maturity-Gated Anti-Forgetting Sampling for Remote-Sensing Object Detection Training
Core Problem: Remote-sensing object detection commonly requires repeated training on high-resolution aerial and satellite imagery, where targets may be small, densely distributed, and surrounded by extensive background.
Key Innovation: The authors propose Maturity-Gated AFSS (MG-AFSS), a detector-maturity activation controller for AFSS. The method accumulates validation mean average precision at an intersection-over-union threshold of 0.50 (mAP@0.50), fits a cumulative saturating curve online, and enables AFSS only when the estimated maturity indicates trustworthy image states.
96. GSANet: Geometric Structure-Aware Siamese Network for 3D Change Detection
Core Problem: Three-dimensional (3D) point cloud change detection is essential for urban monitoring and environmental analysis, yet existing methods mainly rely on point-wise semantic differences and overlook change–unchange boundaries and object edges—key geometric cues for precise localization, especially for subtle or gradual changes.
Key Innovation: To address this, we propose GSANet, a Geometric Structure-Aware Siamese Network that explicitly integrates boundary and edge priors into sampling and feature learning. Second, an Edge-Aware Siamese Network captures robust local shapes by embedding edge priors into feature extraction, incorporating Edge-Aware Adaptive Graph Convolution, Edge-Aware Downsampling, and Cross-Attention Upsampling to maintain structural consistency across temporal branches.
97. Interpretable and uncertainty-aware prediction of fatigue crack growth via a hybrid physics-data fusion architecture
Core Problem: current deep learning models frequently neglect fundamental physics and struggle with complex parameter interactions.
Key Innovation: To address these gaps, this study proposes a hybrid physics-data fusion framework that integrates Physics-Informed Neural Networks (PINN) with Deep Gaussian Processes (DGP) to enhance both robustness and interpretability in crack growth prediction.
98. Aerial Image Gravity Direction Estimation (AGE): Robust single-image estimation by fusing foundation model features with camera geometry
Core Problem: reliable estimation from a single aerial image remains challenging due to the wide range of aerial viewing angles, which often leads to ambiguous geometric cues.
Key Innovation: To resolve geometric ambiguity under diverse viewing configurations, AGE combines semantically rich features from a frozen DINOv2 backbone with camera-aware geometric features derived from known camera intrinsics. Geometric reasoning is guided by supervision from an intermediate Perspective Field, while gravity is predicted as a 3D unit vector for stable learning across aerial viewpoints.
99. HCFNet: Hierarchical and complementary feature embedding for remote sensing change captioning
Core Problem: Remote sensing change captioning (RS-CC) aims to describe changes between bi-temporal remote sensing images in natural language.
Key Innovation: To address these issues, a Hierarchical and Complementary Feature Embedding Network (HCFNet) is proposed. In addition, the Change-Sensitive Metric (CSM) complements traditional n -gram-based metrics with a change-scene penalty term, thereby improving sensitivity to spatial-scale and change-extent mismatches.
100. Digital model reconstruction of shield tunnels based on 3D point cloud
Core Problem: Shield-tunnel condition models need to represent cross-sectional deformation, joint geometry and surface defects directly from measured three-dimensional geometry rather than manual abstraction.
Key Innovation: A key-feature-constrained point-cloud workflow combines global-local ellipse fitting with boundary-constrained defect mapping, achieving about 1 mm geometric deviation, approximately 96% geometric coverage and 85-90% defect coverage without manual reconstruction.
101. Response of pressurized municipal pipelines to tunneling
Core Problem: Tunneling beneath existing pressurized municipal pipelines is becoming increasingly common in urban construction, yet current preliminary evaluation methods often neglect the influence of internal pressure on pressurized pipeline response.
Key Innovation: To address this, an analytical method is proposed based on the classical Euler-Bernoulli beam on a Pasternak foundation model, which innovatively incorporates the coupled effects of tunneling-induced ground deformation and internal pressure-induced stress on the pressurized pipeline response.
102. Dual Scheme Adaptive Slope-Based Filtering of Photogrammetric Point Clouds for Digital Terrain Model Generation
Core Problem: A dual adaptive slope filter separates rough vegetated terrain from flat building-dominated areas before classifying UAV photogrammetric point clouds.
Key Innovation: On a 121 ha survey with 4.15 cm ground sampling and 12 independent GNSS checkpoints, the terrain model reaches 0.148 m vertical RMSE and 0.244 m LE90, although extensive vegetation remains over-smoothed.
103. CogVis: Must Open-Vocabulary Change Detection Perceive the Scene Anew for Every Query?
Core Problem: Open-vocabulary change detection repeatedly entangles temporal evidence with category queries, producing unstable decisions and redundant scene-level computation.
Key Innovation: CogVis separates reusable category-agnostic change perception, query-specific score calibration and region reliability filtering; it leads seven change and building-damage benchmarks while increasing inference throughput by 28.50%.
104. DARAD: Dual Adapters and Ranking-Aware Distillation for Continual Remote Sensing Image-Text Retrieval
Core Problem: continual RS-ITR remains challenging because scale variation and distribution shifts in RS aggravate cross-modal alignment space distortion, making it difficult for existing continual learning (CL) methods to support reliable continual retrieval.
Key Innovation: To address this challenge, we propose DARAD, a dual-adapter and ranking-aware distillation framework that preserves the historical cross-modal ranking structure while learning new visual and textual concepts from evolving archives. Experiments under a multi-stage continual retrieval protocol show that DARAD achieves superior performance over existing CL methods, improving adaptation to newly arrived data while maintaining effectiveness on historical data.
105. Hyperspectral Calibration Detection: A Novel Concept For Change Detection With Unsupervised Incremental Safe Pseudo-Labeling Implementation
Core Problem: in some practical scenarios, such as those requiring immediate detection responses for onboard edge computing, we need to achieve the zero-label requirement as ground-truth labeling would not be available onboard for newly acquired images.
Key Innovation: Based on an iteratively augmented training set that safely collects some unchanged pixel samples, we learn an iteratively refined spectrum calibration function that eventually compensates the variability of acquisition conditions (often observed in bitemporal images), thereby making the changed pixels easily detectable by analyzing the calibrated spectra.
106. Iterate or Widen? When Test-Time Refinement Helps LiDAR Scene Completion: A Controlled Study of Evidence Geometry, Training Coverage, and Compute
Core Problem: Should a completion model spend extra test-time compute by iterating, or spend a similar parameter budget on a wider one-shot predictor?
Key Innovation: These results establish a geometry-conditioned empirical boundary rather than a universal advantage: coherent gaps can justify fixed-depth refinement, broadly thinned evidence is addressed more effectively by training coverage, and spurious evidence requires a different robustness mechanism.
107. MAVISEG: Manifold Propagation and Visual Prototypes for Zero-Shot Open-Vocabulary Segmentation in Diffusion Transformers
Core Problem: Text-to-image diffusion transformers learn about objects and scenes by learning to generate them, making them strong candidates for training-free zero-shot open-vocabulary semantic segmentation.
Key Innovation: The authors present MAVISEG, a training-free refinement layer that recovers these signals. Because its operators consume only a pixel-by-concept score field and a pixel feature space, MAVISEG is capture-agnostic rather than tied to one attribution method.
108. SCI-CLIP: Segment-Centric Inference with Reference Memory for Training-Free Open-Vocabulary Segmentation
Core Problem: Training-free open-vocabulary segmentation remains limited by a missing inference abstraction.
Key Innovation: The authors present SCI-CLIP, a segment-centric inference framework built around the principle that the same region abstraction should organize all stages of dense open-vocabulary prediction.
109. Overcoming Attention Drift: Homogeneity-Heterogeneity Guided Feature Aggregation for Low-Light Remote Sensing Image Enhancement
Core Problem: under severe noise and illumination corruption, existing methods suffer from attention drift, erroneously aggregating features across distinct physical boundaries and causing severe structural blurring and color distortion.
Key Innovation: To address this, we propose HALO, a dual-prior-driven enhancement framework that formulates enhancement as a guided feature aggregation problem driven by foundation model priors.
110. SkillTFM: Gated Skill Evolution for Training-Free Adaptation of Tabular Foundation Models
Core Problem: Tabular data are ubiquitous in real-world applications and are crucial for data-driven prediction and decision-making across science, industry, finance, healthcare, and public services.
Key Innovation: Tabular foundation models (TFMs) have emerged as a promising paradigm for general-purpose tabular learning, offering reusable predictors across diverse datasets and substantially reducing the need for task-specific training, tuning, and model development.
111. Do Tabular Foundation Models Agree with Themselves?
Core Problem: Autoregressive use of tabular foundation models assumes that separately predicted conditionals define a coherent joint distribution, an assumption that had not been tested on real data.
Key Innovation: The study formalizes marginalization and factorization consistency tests and finds that every evaluated tabular foundation model violates both requirements across classification and regression datasets.
112. Universal Concept Disruption for SAM3 Image Segmentation
Core Problem: SAM3 extends promptable segmentation from geometry-driven mask prediction to open-vocabulary concept segmentation, where a text-conditioned grounding model decides whether a concept is present and segments all matching instances.
Key Innovation: While this presence-gated design improves concept-level prediction, its adversarial robustness remains unexplored.
113. UQ-Loc: Uncertainty-Aware LiDAR Scene Coordinate Regression
Core Problem: existing methods produce deterministic predictions, discarding aleatoric uncertainty that could improve robustness and downstream decision-making.
Key Innovation: The authors present UQ-Loc, which extends the LightLoc architecture with an anisotropic Gaussian covariance head that predicts a full 3x3 positive-definite covariance matrix per voxel.
114. Task-Guided CycleGAN for Domain-Adapted Coastal Boundary Segmentation in Remote Sensing Images
Core Problem: Coastal-boundary segmenters generalize poorly across sensors and regions, while conventional fine-tuning changes the learned boundary representation and requires target labels.
Key Innovation: Two task-guided CycleGAN variants translate six source datasets into the target style while keeping SegFormer fixed; cross-domain gains are largest for spectrally distinct near-infrared sources.
115. MPC-DETR: A Multi-Scale Patch Context Transformer for Small Object Detection in UAV Imagery
Core Problem: Small-object detection in unmanned aerial vehicle (UAV) imagery remains challenging because target objects often occupy only a few pixels, exhibit weak feature responses, and are easily obscured by complex backgrounds.
Key Innovation: To overcome these limitations and enhance the detection accuracy in challenging UAV settings, this paper proposes MPC-DETR, a Multi-scale Patch Context Transformer that is based on RT-DETR. To begin with, a Local-Global Attention Fusion Module (LGAF) is proposed to capture fine-grained local features and long-range semantic relations of small objects.
116. Quality Assessment and Observation Error Estimation of Tianmu-1 GNSS Radio Occultation Bending Angle and Refractivity Retrievals
Core Problem: As China’s first commercial low Earth orbit meteorological constellation, Tianmu-1 (TM) radio occultation (RO) provides high-density observations, with approximately 30,000 profiles per day globally, offering new opportunities for research in observationally sparse regions.
Key Innovation: The study quantitatively characterizes the accuracy, precision, and error structure of TM RO bending angle and refractivity, providing a quantitative basis for future quality control and observation error specification in numerical weather prediction data assimilation.
117. Multiscale investigation on the self-sealing behaviour of granular bentonite
Core Problem: The multiscale processes by which granular bentonite closes technological gaps, and the consequences for water and gas transport in engineered barriers, remain underexplored.
Key Innovation: X-ray microtomography and mercury-intrusion measurements show that initial compaction controls matrix structure and sealing mode: dry-state granules rearrange into the gap, whereas wet-state aggregates require high-stress saturation for extensive closure.
118. Multiscale damage evolution mechanism of Longmaxi shale subjected to supercritical CO2 soaking
Core Problem: The study systematically examines the time-dependent deterioration and failure evolution of Longmaxi Formation shale under supercritical CO₂ (SCCO2) exposure for different soaking durations (0, 10, and 30 days).
Key Innovation: Based on the equivalent stress hypothesis and energy dissipation principles, a combined damage variable (D comb) was derived to couple the initial chemical-induced pre-weakening with loading-induced evolutionary damage. Compared with the unsoaked group, the peak load-bearing capacity decreased by 17.9% after 10 days and by 21.9% after 30 days of soaking.
119. In-situ stress controls on reservoir quality and permeability anisotropy: A case study of the continental shale oil system in the southern Songliao Basin
Core Problem: Heterogeneous in-situ stress can reorganize fractured-rock permeability and reservoir quality, yet its controls across structural domains remain insufficiently quantified.
Key Innovation: Kaiser tests, pressure-dependent permeability experiments and field observations separate three geomechanical domains in the southern Songliao Basin and show stress-dependent anisotropy, with the central extensional domain providing the most favourable hydraulic conditions.
120. Global cloud phase monitoring from FY-3G multi-angle polarimetric observations using the new MAPLE-CP framework with independent multi-site lidar validation
Core Problem: Global cloud-top phase monitoring is constrained by an observing gap: passive imagers provide broad coverage but limited vertical phase information, whereas active sensors provide stronger physical constraints only along narrow tracks.
Key Innovation: Here we develop MAPLE-CP, a Multi-Angle Polarimetric Latent-Enhancement framework for FY-3G PMAI cloud-top phase retrieval. MAPLE-CP links shortwave-infrared multi-angle polarimetric scattering information, DQ-1 ACDL cloud-top lidar labels, physics-guided feature screening, compact latent augmentation, and independent lidar validation.
121. Reducing grid artifacts in air temperature downscaling with a Volume-Preserving interpolation preprocessing step
Core Problem: coarse baseline data often exhibit sharp discontinuities at grid boundaries, which propagate into downscaled outputs as unnatural artifacts and spatial discretization errors.
Key Innovation: Five widely used machine learning based downscaling algorithms were adopted to evaluate the general applicability of our proposed approach. Furthermore, to improve preprocessing efficiency, we developed a Gaussian filtering variant of pycnophylactic interpolation and compared it with the conventional method.
122. Cumulative and nonlinear reshaping of slope topography under repeated manual contour ploughing on the Chinese Loess Plateau
Core Problem: Tillage erosion is a major driver of topographic change on sloping farmland, yet the long-term geomorphic effects of intensive manual contour ploughing remain poorly quantified.
Key Innovation: Here, a full-factorial runoff-plot experiment combining six levels of slope gradients (5.24%, 8.75%, 17.63%, 26.79%, 36.40%, and 46.63%) and six tillage-pass treatments (1, 6, 12, 18, 24, and 30 passes) was used to examine how repeated manual contour ploughing reshapes slope topography on the Chinese Loess Plateau.
123. Cyclic shear response of asphalt concrete–sand gravel interface using FDM–DEM coupling
Core Problem: Cyclic load transfer across asphalt-concrete and sand-gravel interfaces is poorly understood despite its control on deformation compatibility in core rockfill dams.
Key Innovation: Coupled finite-difference and discrete-element simulations link early-cycle strength loss and shear contraction to particle rearrangement, force-chain reconstruction and evolving contact anisotropy; higher normal stress increases strength and suppresses stiffness degradation.
124. DEM investigation of soil arching effect induced by vertical unloading under high geostress
Core Problem: To reveal the evolution mechanism of the soil arching effect induced by vertical unloading under high geostress, a validated discrete element modelling framework was developed.
Key Innovation: The framework was established based on trapdoor model tests conducted at simulated burial depths of 10-90 m using a self-developed high geostress trapdoor apparatus, and was validated against both the ground reaction curve (GRC) and the displacement field.
125. DEM-enabled exploration of spatial inhomogeneities during the cyclic torsional shear of sand
Core Problem: Cyclic torsional shear of a poorly graded sand was simulated with LIGGGHTS using three particle shape representations: spheres with rolling friction, ellipsoids and multi-sphere clumps.
Key Innovation: The torsional shear apparatus was simulated in DEM with two meshes to represent the top and bottom end-caps. The bulk behaviour after calibration was captured reasonably although the damping was overestimated at higher strains.
126. Probabilistic models for free-fall penetrometer interpretation in clay
Core Problem: Dynamic penetration and strain-rate effects make free-fall penetrometer estimates of clay resistance both uncertain and heteroscedastic.
Key Innovation: Bayesian calibration and variational Bayesian neural networks quantify parameter, aleatoric and epistemic uncertainty; the neural models reach about 0.9 R-squared compared with 0.76 for calibrated semi-empirical equations.
127. Dynamic responses of silty clay under repeated high-amplitude loading and low confining pressure
Core Problem: Silty clay in unpaved roads experiences low-cycle wheel loading at high dynamic stress and low confinement, but its transition from stable deformation to incremental collapse remains insufficiently constrained.
Key Innovation: Cyclic triaxial tests separate plastic shakedown, plastic creep and incremental collapse within 50 cycles; an initial-strain indicator and fitted modulus and strain models identify critical stress states across water-content and confinement conditions.
128. How exposed root distribution patterns regulate overland flow hydrodynamics on slope segments: a simulated scouring study
Core Problem: The exposure of plant roots caused by soil erosion significantly alters surface microtopography and runoff pathways; however, the specific mechanisms by which it affects hydrodynamic processes remain not fully understood.
Key Innovation: The study investigated how slope segment, exposed root distribution pattern, and runoff discharge influence the hydraulic characteristics of overland flow by simulated scouring experiments. Results revealed that transverse and cross-slope root arrangements effectively reduced flow velocity and erosive energy, whereas parallel arrangement had a minimal effect.
129. Nuclear magnetic resonance relaxometry for pore size distribution in porous media: Accounting for diffusional coupling
Core Problem: Nuclear-magnetic-resonance pore-size estimates become biased when water diffuses between connected pore classes during relaxation.
Key Innovation: A binary-pore diffusion model, compared with mercury-intrusion measurements for four clays, shows progressive merging of smalland large-pore peaks and substantial size and volume errors once the coupling coefficient exceeds 40.
130. Laser-cutting of fracture network in binder jetting 3D-printed rock mass
Core Problem: Laboratory models of fractured rock require complex discontinuity networks with enough geometric fidelity to reproduce natural structural controls.
Key Innovation: Non-contact laser cutting is combined with binder-jet 3D printing to fabricate networks containing up to 248 fractures, with less than 7% fractal-dimension deviation and less than 3% length error from the designs.
131. Mode I-II fracture behaviors of high-temperature treated granite: Insights from experimental and peridynamic simulations
Core Problem: the mode I-II fracture mechanism of crystalline rocks under high-temperature conditions remains unclear.
Key Innovation: The study examined cracked straight-through Brazilian disc (CSTBD) granite by combining digital image correlation (DIC) and acoustic emission (AE) techniques under varying temperatures and fissure inclination angles.
132. Shearing responses of bolted rock joints subjected to normal-tangential cyclic loading
Core Problem: Based on a self-developed high-strength and high-toughness negative Poisson's ratio (NPR) bolt, dynamic shear tests were conducted on bolted jointed rock masses (BJRM) under coupled sinusoidal cyclic normal load and cyclic shear displacement.
Key Innovation: The influences of initial normal load (F s), normal dynamic load frequency (f v), and joint roughness coefficient (JRC) on the shear behavior of bolted joints were investigated. The NPR bolt significantly enhanced the shear resistance of jointed rock masses under dynamic cyclic shear, and the NPR anchoring system improved the peak shear force (F hmax) by over 30.75%.
133. Investigation of damage evolution and heterogeneity in intact Wombeyan marble using a continuum grain-based model
Core Problem: Continuum grain-based models must reproduce grain-scale stress and strain heterogeneity without the computational cost of discontinuum simulations, but crack connectivity can distort their high-load response.
Key Innovation: A calibrated marble model combines continuum grains with digital-image-correlation strain reconstruction; increasing crack interconnectivity restores experimental strain-heterogeneity trends while retaining continuum-scale efficiency.