TerraMosaic Daily Digest: June 15, 2026
Daily Summary
The June 15 issue is led by papers that connect terrain instability to the wider systems in which hazards propagate. A Hindu Kush study treats earthquakes, floods, landslides, debris flows, and glacial lake outburst floods as interacting risk pathways rather than independent layers. Two Engineering Geology papers strengthen the subsurface side of landslide assessment: one converts seismic refraction tomography into volumetric 2D-3D RQD fields for CPEC slope terrain, while another tests unmanned aerial transient electromagnetics against electrical resistivity tomography for rapid slip-surface investigation. The Landslides paper on the Ohya landslide scar adds a complementary surface archive, using UAV multispectral indices to reconstruct debris-flow histories where conventional records are incomplete.
The strongest operational thread is the movement from mapped hazard to response-ready evidence. MAS-LAND combines transformer-based landslide detection, infrastructure exposure assessment, and automated reporting inside a multi-agent response workflow. A Bulletin of Earthquake Engineering paper quantifies how earthquake-induced landslides affect urban and major-road resilience, while a non-hydrostatic landslide-wave model follows subaerial granular failures into water-body impacts. A Remote Sensing case study adds mining-induced deformation to this operational cluster by linking InSAR monitoring with rock-movement theory for steep mountainous slopes. Earthquake and flood papers extend the same logic to liquefiable transit networks, bridge response, liquefiable sloping ground, urban flood-resilient design, coastal water-level extremes, and early seismic deformation monitoring.
AI and remote-sensing work is increasingly evaluated by whether it respects geometry, physics, and mission constraints. GeoDisaster benchmarks tool-grounded disaster geo-intelligence across multi-hazard reasoning, building damage, flood routing, and SAR flood monitoring. Science Advances papers on sea-level attribution, ENSO predictability, and a data-driven global ocean model show that foundation-style Earth-system models are becoming relevant to hazard attribution and long-range forecasting. ArXiv studies on SAR-optical fusion, frequency-domain disaster assessment, causal tipping-point diagnostics, physics-constrained weather models, and RPC-consistent satellite feature evaluation push the field toward models whose outputs remain interpretable under cloud cover, sparse data, complex viewing geometry, and nonstationary extremes. A multisource assessment of rainfall-extreme-induced aboveground biomass loss adds a complementary impact-mapping perspective for events that combine flooding, landslides, and erosion.
Key Trends
Five movements define the issue: coupled multi-hazard risk, subsurface landslide evidence, workflow-aware geohazard AI, physics-constrained remote sensing, and rapid probabilistic earthquake response.
- Multi-hazard risk is being modeled as interaction rather than overlay: The Hindu Kush, earthquake-induced road resilience, landslide-generated waves, flood-resilient neighborhoods, coastal water-level extremes, and liquefiable transit-network papers all treat hazard consequences as coupled chains.
- Landslide investigation is moving below the surface: Seismic-refraction-to-RQD modeling, unmanned aerial transient electromagnetics, electrical resistivity validation, UAV multispectral debris-flow reconstruction, and landslide susceptibility sampling all expand the evidence base beyond visible scarps.
- Operational geohazard AI is becoming workflow-aware: MAS-LAND and GeoDisaster both evaluate whether models can support detection, exposure analysis, routing, reporting, and evidence-backed decisions instead of isolated image classification.
- Remote-sensing methods are being constrained by physics and geometry: SAR-optical fusion under clouds, RPC-consistent feature evaluation, mining-induced InSAR slope monitoring, rainfall-extreme impact mapping, and physics-constrained weather forecasting emphasize observation reliability under real sensing conditions.
- Earthquake resilience papers are converging on rapid, probabilistic damage states: Liquefiable rail networks, pile loads in sloping ground, bridge post-earthquake response, mainshock-aftershock fragility, floor response spectra, and early GNSS deformation monitoring all convert seismic forcing into actionable system response metrics.
Selected Papers
The selected papers emphasize multi-hazard mountain risk, landslide subsurface investigation, debris-flow history reconstruction, landslide-generated waves, operational landslide response, earthquake-induced landslide road resilience, mining-induced InSAR slope monitoring, coastal flood attribution, AI Earth-system modeling, cloud-robust remote sensing, liquefaction, seismic infrastructure response, seepage erosion, and geotechnical material degradation. This issue contains 43 selected papers from 2098 papers analyzed.
1. Integrated multi-hazard vulnerability and risk assessment in the Hindu Kush Mountains of Northern Pakistan: Challenges and recommendations
Core Problem: Chitral in the Hindu Kush is exposed to earthquakes, floods, landslides, debris flows, and glacial lake outburst floods, yet integrated risk assessments often underrepresent how these hazards interact.
Key Innovation: Combines open-source remote-sensing data, field observations, hazard weighting, exposure, and vulnerability information to build a district-scale multi-hazard vulnerability and risk assessment.
2. Volumetric 2D-3D RQD modeling from seismic refraction tomography for landslide susceptibility assessment in complex geological terrain
Core Problem: Landslide-prone infrastructure corridors often lack continuous subsurface rock-quality information because surface mapping and sparse boreholes undersample complex terrain.
Key Innovation: Develops a site-calibrated seismic refraction tomography to RQD workflow that converts velocity structure into continuous 2D and 3D rock-quality models for susceptibility assessment.
3. Debris-flow historical assessment using UAV-mounted multispectral cameras at Ohya landslide scar, Japan
Core Problem: Debris-flow histories are often incomplete where field records are sparse, limiting countermeasure planning in active landslide scars.
Key Innovation: Uses UAV-mounted multispectral sensing, red-edge and near-infrared reflectance, NDVI, and red-edge indices to reconstruct debris-flow activity at the Ohya landslide scar.
4. Rapid investigation of a complex-terrain landslide using the unmanned aerial transient electromagnetic method: Validation against electrical resistivity tomography
Core Problem: Complex mountainous landslides require fast slip-surface and instability-zone mapping, but contact-based geophysical surveys are slow and terrain-limited.
Key Innovation: Applies unmanned aerial transient electromagnetics to the Huadian landslide and validates the airborne interpretation against electrical resistivity tomography, observations, and meteorological context.
5. An efficient non-hydrostatic numerical model for subaerial granular landslide-generated waves
Core Problem: Subaerial granular landslides can generate dispersive waves that are poorly captured by purely hydrostatic depth-integrated models.
Key Innovation: Couples a Cartesian Savage-Hutter granular-flow model with a two-layer non-hydrostatic water model to reproduce leading waves and dispersive wave trains in benchmark experiments.
6. MAS-LAND: A Multi-Agent System for Landslide detection and rapid response
Core Problem: Post-event landslide mapping often stops at detection accuracy and does not connect mapping outputs to exposure assessment and civil-protection reporting.
Key Innovation: Builds a multi-agent, LLM-enhanced workflow for post-event landslide detection, infrastructure exposure analysis, and automated response reports using very-high-resolution imagery.
7. Landslide susceptibility mapping based on Inter.iamb-Tabu algorithm considering non-landslide sampling
Core Problem: Landslide susceptibility models remain sensitive to model selection and to how non-landslide samples are defined.
Key Innovation: Combines conditioning-factor screening, non-landslide sampling design, and the Inter.iamb-Tabu algorithm to build a susceptibility framework for Gangu County, China.
8. Mining-Induced Deformation and Slope Stability in Steep Mountainous Areas Based on InSAR Monitoring and Rock Movement Theory: A Case Study from Southwestern China
Core Problem: Steep mountainous mining districts require deformation monitoring that can connect surface displacement to slope-stability interpretation.
Key Innovation: Uses InSAR monitoring together with rock-movement theory in a Southwestern China case study to evaluate mining-induced deformation and slope stability.
9. Resilience assessment of the urban and major roads endangered by earthquake-induced landslide
Core Problem: Transportation lifelines endangered by earthquake-induced landslides require performance-based resilience metrics that link slope response to road service loss.
Key Innovation: Uses PLAXIS 2D slope models, recorded ground motions, multiple slope geometries, and seismic-response simulations to quantify resilience of urban and major roads.
10. GeoDisaster: Benchmarking Orchestrated Agents for Operational Disaster Geo-Intelligence
Core Problem: Remote-sensing vision-language models are rarely tested on tool-grounded disaster workflows that require structured, evidence-backed geospatial decisions.
Key Innovation: Introduces a benchmark with verified tasks spanning multi-hazard analysis, building-damage assessment, flood-safe routing, SAR flood monitoring, exposure estimation, and diagnostic reporting.
11. Human-caused sea level rise drives 21st-century worldwide water level extremes
Core Problem: Local flood-risk decisions need quantitative attribution of how anthropogenic sea-level rise changes daily extreme water-level exceedances.
Key Innovation: Quantifies human-caused sea-level rise at 97% of 519 tide gauges and attributes most observed daily extreme water-level exceedances over 2000-2018 to that signal.
12. Data-driven global ocean model resolving atmospherically forced ocean dynamics
Core Problem: Extending AI weather prediction toward climate-relevant hazards requires ocean models that capture three-dimensional responses to atmospheric forcing.
Key Innovation: Presents KIST-Ocean, a deep-learning global ocean general circulation model that reproduces Kelvin waves, Rossby waves, and wind-stress-curl-driven vertical motions.
13. Seismic site effects and liquefaction susceptibility in Larache (Morocco) based on ambient vibrations and shear strain
Core Problem: Urban earthquake vulnerability depends on spatially variable site effects and liquefaction susceptibility under complex near-surface geology.
Key Innovation: Combines HVSR, MASW, boreholes, probabilistic seismic hazard assessment, and shear-strain analysis to map site response and ground-deformation susceptibility in Larache.
14. Seismic response of an embankment slope reinforced with rubble mound: centrifuge model testing and analytical validation
Core Problem: Embankment topography can amplify earthquake motion and destabilize structures placed on or near the slope.
Key Innovation: Uses dynamic centrifuge tests and numerical modeling to quantify how rubble-mound reinforcement reduces acceleration response and increases slope factor of safety.
15. Full-process seismic damage assessment for urban underground rail transit networks in liquefiable sites
Core Problem: Regional rail-transit networks in liquefiable sites require network-scale seismic damage assessment without sacrificing the physics of ground-motion and soil response.
Key Innovation: Integrates source-to-site simulation, boundary-surface constitutive modeling, liquefaction-potential mapping, and generalized response displacement analysis for full-process damage assessment.
16. Rapid assessment of bridge post-earthquake response using unsupervised clustering and probabilistic machine learning under large-scale Kik-net data
Core Problem: Post-earthquake bridge response assessment is constrained by scarce labeled ground-motion and structural-response data.
Key Innovation: Uses K-means clustering on large-scale Kik-net records and probabilistic machine learning to create a rapid bridge seismic response assessment framework.
17. An integrated framework for decoupling geodetic parameters in early seismic deformation monitoring: a case study of the 2011 MW 9.0 Tohoku earthquake
Core Problem: Real-time GNSS deformation monitoring can produce false coseismic signals when ambiguities, atmospheric delays, clocks, and position estimates remain coupled.
Key Innovation: Separates geodetic parameters through a model-driven decoupling framework and reduces early coseismic deformation errors toward millimetre-scale accuracy.
18. Deep learning reveals enhanced ENSO predictability under historical anthropogenic forcing
Core Problem: ENSO predictability controls seasonal hazard outlooks, but the effect of anthropogenic forcing on that predictability remains uncertain.
Key Innovation: Applies CNN-based leave-one-out modeling to CMIP6 simulations and links enhanced ENSO predictability to anthropogenic changes in ocean-atmosphere feedbacks.
19. Integration of satellite monitoring and 3D modelling approach for multi-scale analysis of existing tunnel infrastructures
Core Problem: Existing tunnels are exposed to surrounding ground deformation and internal structural degradation, but monitoring often separates regional and asset-scale evidence.
Key Innovation: Combines InSAR ground-deformation detection with high-resolution 3D tunnel-lining reconstruction to support multi-scale infrastructure condition assessment.
20. Designing flood-resilient urban neighborhoods by integrating hydrodynamic modeling, blue-green infrastructure, and community engagement
Core Problem: Flood-vulnerable neighborhoods require design strategies that connect hydrodynamic evidence with implementable urban adaptation measures.
Key Innovation: Integrates flood modeling, blue-green infrastructure design, and community engagement to formulate flood-resilient neighborhood strategies in Sarpol-e Zahab, Iran.
21. Advancing the measurement of social vulnerability to earthquakes at subnational levels of geography: a contextually valid and robust framework
Core Problem: Composite social-vulnerability metrics are often weakly validated and insufficiently tailored to earthquake contexts.
Key Innovation: Develops a context-specific framework for earthquake social vulnerability at subnational scales, emphasizing indicator validity and robustness.
22. Centrifuge model tests for distinguishing kinematic and inertial loads on single piles in liquefiable sloping ground
Core Problem: Pile design in liquefiable sloping ground remains uncertain because guidelines differ on how to combine kinematic and inertial loads.
Key Innovation: Uses centrifuge shaking-table tests on single piles with and without superstructure mass to separate kinematic and inertial load phases during lateral spreading.
23. Prediction of the Kamchatka July 29, 2025, earthquake by the evolution of low-magnitude seismicity recovered using waveform cross-correlation at IMS seismic arrays
Core Problem: Low-magnitude seismicity preceding major earthquakes can fall below standard detection thresholds, limiting analysis of pre-rupture evolution.
Key Innovation: Uses waveform cross-correlation at IMS seismic arrays to recover weak events and analyze recurrence-curve evolution before the July 2025 Kamchatka earthquake.
24. Joint Analysis of Shannon and Tsallis Entropy and GRACE-FO driven Equivalent Water Height Anomalies for Pre- and Post-Rupture Monitoring: An Example of the 2023 Mw = 7.8 Kahramanmaras Earthquake, Turkiye
Core Problem: Fault-system state changes before and after large earthquakes are difficult to isolate from seismic catalogs alone.
Key Innovation: Combines 25 years of seismicity, GRACE-FO equivalent water-height anomalies, and Shannon and Tsallis entropy to track pre-, co-, and post-rupture fault-system complexity.
25. SpatioTemporal Causal Network Diagnostics for Geographic Tipping Point Early Warning
Core Problem: Classical spatial early-warning indicators can dilute localized tipping signals and mis-handle correlated noise.
Key Innovation: Represents geographic fields as directed causal networks, estimates local recovery rates, and identifies vulnerable subnetworks in SST and AMOC benchmarks.
26. Heterogeneous SAR-optical fusion for near-real-time land use and land cover mapping under cloud contamination: A novel framework and global benchmark dataset
Core Problem: Cloud and shadow contamination reduce the reliability of near-real-time optical mapping during hazard response.
Key Innovation: Introduces CloudLULC-Net and a large SAR-optical benchmark, using optical reliability modulation and adaptive SAR-optical aggregation for cloudy Sentinel-2 scenes.
27. Bridging Spatial And Frequency Views For Disaster Assessment: Benefits And Limitations
Core Problem: Building-damage classifiers based only on spatial features may miss frequency-domain cues associated with debris, collapse texture, and structural disruption.
Key Innovation: Compares spatial, frequency, and dual-domain EfficientNet approaches for xView2 multi-class building-damage classification under matched training conditions.
28. Physics-Constrained Neural Networks for Improved Short-Term Weather Forecasting: A Case Study over the South Pacific
Core Problem: Short-term weather forecasts used in hazard anticipation need neural accuracy without losing physical stability.
Key Innovation: Adds WENO-5 numerics, a unified autoregressive hybrid block, and physics-informed neural backbones to improve 1-12 h WeatherBench South Pacific forecasts.
29. Stratigraphic transformer: Encoding sparse boreholes to learn stratigraphic structure without prior geological knowledge
Core Problem: Sparse boreholes make it difficult to infer locally varying stratigraphic boundaries and termination features in unexplored areas.
Key Innovation: Uses a transformer architecture to learn global and local dependencies from sparse borehole labels while quantifying uncertainty in reconstructed stratigraphy.
30. Multi-factor influences on seepage erosion in gap-graded soils induced by joint leakage: a model test
Core Problem: Joint leakage can trigger seepage erosion in sandy ground, threatening underground structures through evolving erosion zones and pore-pressure changes.
Key Innovation: Uses physical model tests to separate hydraulic-gradient, earth-pressure, and joint-size effects on erosion morphology, particle loss, pore pressure, and failure stages.
31. The influence of porosity on the damage induced by freeze-thaw cycles on fracture toughness
Core Problem: Freeze-thaw cycles can reduce rock-bridge strength and fracture toughness in periglacial slopes, but low-porosity rock response remains poorly constrained.
Key Innovation: Tests contrasting lithologies under freeze-thaw cycling to quantify changes in Mode I fracture toughness and fracture-process behavior.
32. New insights into the reinforcement and failure mechanisms of biocemented fracture zone
Core Problem: Fault fracture zones in slopes, tunnels, and dam foundations combine low strength, high permeability, and loose structure.
Key Innovation: Uses MICP, computed tomography, SEM, and image processing to analyze calcium-carbonate deposition, seepage-channel sealing, and macro-meso failure mechanisms.
33. A framework for assessing rainfall extremes induced aboveground biomass loss from multisource remote sensing data
Core Problem: Extreme rainfall can combine flooding, landslides, and erosion with vegetation loss, but event-scale aboveground-biomass impacts remain difficult to quantify from coarse products.
Key Innovation: Integrates high-resolution change detection with XGBoost downscaling of biomass products to estimate event-scale aboveground biomass loss from multisource remote sensing data.
34. The effects of soluble and insoluble salts on the small to large strain shear behaviour of reconstituted clayey loess
Core Problem: Salt content changes can alter the mechanical behavior of loess used in foundation and slope stability assessments.
Key Innovation: Uses triaxial, bender-element, and microstructural tests to compare soluble NaCl and insoluble CaCO3 effects from small to large strain.
35. Seismic Velocity Variations Illustrate Hydrological and Ecohydrological Processes in a Mountain Watershed
Core Problem: Hydrological states that influence slope stability are difficult to observe continuously across hillslope transects.
Key Innovation: Applies passive seismic interferometry to relate seismic velocity variations to groundwater, soil moisture, precipitation, and aspect-controlled ecohydrological behavior.
36. Fragility of irregular RC buildings under mainshock-aftershock sequences
Core Problem: Buildings damaged by a mainshock can collapse under weaker aftershocks, yet single-event fragility models underrepresent this sequence effect.
Key Innovation: Runs incremental dynamic analyses of plan- and vertically irregular RC frames under mainshock-aftershock sequences and compares regularized configurations.
37. TFRSNet and OFRSNet: A Novel Approach for Predicting Floor Response Spectra of Isolated Structures Based on Deep Learning
Core Problem: Floor response spectra for nonstructural components are costly to compute repeatedly for isolated structures with nonlinear response.
Key Innovation: Develops deep-learning models to predict floor response spectra efficiently for isolated structures, reducing dependence on repeated time-history analysis.
38. Correlations of horizontal and vertical spectral accelerations for the Chilean subduction zone
Core Problem: Vertical and horizontal spectral accelerations are correlated, but regional correlation models remain limited for subduction-zone applications.
Key Innovation: Builds parametric correlation models for Chilean subduction-zone records and tests magnitude, distance, site-condition, and model-selection effects.
39. Probabilistic seismic demand model based on period shift and optimal intensity measure selection
Core Problem: Stiffness degradation during earthquakes shifts structural periods, but this diagnostic variable is underused in seismic demand modeling.
Key Innovation: Builds probabilistic seismic demand models using period shift as an engineering demand parameter and evaluates optimal intensity measures under uncertainty.
40. Dynamic response and improved safety prediction method for large diameter pipelines with different shapes subjected to close-proximity tunnel blasting vibration
Core Problem: Large-diameter pipelines with aging cracks may fail when exposed to close-proximity blasting vibrations during tunnel construction.
Key Innovation: Uses explicit dynamic finite-element modeling to compare vibration response and failure locations across circular, square, and horseshoe-shaped pipelines.
41. Geometric Consistency Protocol for Foundation Model Features in Multi-View Satellite Imagery
Core Problem: Remote-sensing foundation features can appear semantically consistent while failing geometric localization under multi-view satellite imaging constraints.
Key Innovation: Introduces an RPC-consistent evaluation protocol using projected 3D consistency and geometry-constrained dense matching to benchmark foundation features.
42. How Sparse and How Noisy? Systematic Benchmarking of Inverse Physics-Informed Neural Networks for Manning Friction Estimation in Shallow Water Equations
Core Problem: PINN-based flood and shallow-water modeling needs clearer limits under sparse observations and noisy data.
Key Innovation: Benchmarks inverse PINN recovery of Manning friction under controlled sparsity, noise, and observation-variable settings in one- and two-dimensional shallow-water cases.
43. ED3R: Energy-Aware Distributed Disaster Detection Enabled by Cooperative Robotic Agents
Core Problem: Robotic disaster monitoring must balance detection confidence against energy, time, uncertainty, and obstacle constraints.
Key Innovation: Introduces a cooperative robot-controller framework for wildfire detection with energy-aware decision-making, adaptive early completion, and distributed neural prediction.