TerraMosaic Daily Digest: June 21, 2026
Daily Summary
The June 21 literature is unusually strong on landslide process, monitoring, and cascading hazard mechanics. Two Landslides papers anchor the issue: one reconstructs earthquake-induced rockslide damming and lake formation in Mongolia's Gobi-Altay Range, and the other dissects the March 2025 Poggio Baldi sequence with multi-sensor evidence. Engineering Geology adds the mechanics needed to interpret such failures, from rate-dependent residual shear strength to a unified SPH treatment of landslide-damming-breaching-flood cascades, rainfall-reactivated loess fissures, and InSAR-supported hazard prediction in mining backfill and dumping zones.
A second cluster moves from single-hazard maps toward coupled exposure and attribution. Flood papers combine livelihood vulnerability, dam-release inundation, valley-floor geomorphometry, Bayesian disaster-loss attribution, and flash-flood reconstruction. Marine and cryospheric hazards appear through meteotsunami reviews, unusual tidal event forensics, snow-drift railway protection, and subsea cable sensing for ocean and earthquake monitoring. Together, the issue treats hydrometeorological forcing, terrain response, infrastructure exposure, and uncertainty as interacting parts of the same risk system.
The methods signal is practical rather than decorative. Deep learning is used where it supplies missing hazard observables: loess sinkhole mapping, UAV slope deformation monitoring, aerial-image segmentation, disaster damage assessment, and InSAR image matching. Physics-informed neural networks, multi-physics debris-flow impact models, machine-learning bearing-capacity estimators, and geophysical datasets extend the computational layer without detaching it from process. The strongest papers therefore push TerraMosaic-relevant science toward auditable hazard evidence: measured deformation, explicit mechanics, validated exposure, and reusable data streams.
Key Trends
Five movements define this issue: sequence-aware landslide mechanics, end-to-end cascading-hazard simulation, exposure-aware hazard mapping, physical remote-sensing observables, and geophysically constrained AI.
- Landslide science is becoming sequence-aware: Rockslide-dammed lakes, rainfall-triggered Poggio Baldi failures, residual-strength rate effects, and RSF slope timing all frame landslides as evolving mechanical sequences rather than static susceptibility labels.
- Cascading geohazards are being simulated end to end: The LDBF SPH framework, debris-flow barrier coupling, flash-flood reconstruction, and dam-release flood prediction show a shift toward models that preserve process continuity across failure, blockage, breach, flow, and impact.
- Hazard mapping is absorbing exposure and attribution: Flood vulnerability, Bayesian disaster-loss attribution, soil-erosion hotspots, snow-drift railway experiments, and seismic rock-slope portal analysis connect hazard intensity to the assets, routes, and populations at risk.
- Remote sensing is moving closer to physical observables: InSAR settlement correction, UAV deformation monitoring, DEM fitness testing, LiDAR photogrammetry, and morpho-phase image matching turn remote-sensing products into measurements of displacement, shape, and geomorphic reliability.
- AI methods are strongest when tied to geophysical constraints: Multimodal U-Nets, physics-informed tomography, multi-agent damage assessment, and VFM distillation are most relevant where they expose uncertainty, respect terrain or sensor geometry, or improve disaster-specific evidence extraction.
Selected Papers
The selected papers emphasize earthquake-induced rockslide dams, rainfall-triggered landslide sequences, residual shear strength, landslide-damming-breaching-flood simulation, loess sinkholes and fissures, InSAR-supported mining hazards, debris-flow barriers, flash-flood reconstruction, flood vulnerability and loss attribution, meteotsunamis, snow-drift hazards, seismic tomography, dense-array data, geophysical monitoring, and disaster-oriented remote-sensing AI. This issue contains 49 selected papers from 3123 papers analyzed.
1. Geomorphic characteristics of lake depressions in the Gobi-Altay Range, Mongolia: Impact of an earthquake-induced rockslide
Core Problem: Large coseismic rockslides can reorganize valley hydrology, but their damming products and lake depressions are often preserved only as fragmented geomorphic evidence.
Key Innovation: Combines remote sensing, morphometry, and field measurements to reconstruct three rockslide-dammed lakes formed by the 1957 Mw 8.1 Gobi-Altay earthquake.
2. Multi-sensor anatomy of the March 2025 landslide sequence in Poggio Baldi (Northern Apennines, Italy)
Core Problem: Sequential landslide failures can involve different materials and failure styles over only a few days, making single-sensor interpretation incomplete.
Key Innovation: Documents an earth-flow to rock-slide sequence at Poggio Baldi using rainfall records and multi-sensor observations to resolve timing, geometry, and triggering.
3. A unified 3D SPH framework for simulating the landslide-damming-breaching-flood cascading hazards
Core Problem: Landslide-dam failures involve coupled mass movement, dam formation, breach erosion, and downstream flooding that are usually modeled with disconnected tools.
Key Innovation: Develops a Riemann-based fluid-soil coupled SPH framework that simulates the complete LDBF chain within one three-dimensional formulation.
4. Rate dependence of residual shear strength: Roles of clay fraction and uncertainties
Core Problem: Runout and slow-creep predictions remain uncertain because rate-weakening and rate-strengthening friction are rarely evaluated objectively across soil mixtures.
Key Innovation: Uses ring-shear experiments on clay-sand mixtures to quantify how clay fraction and loading conditions control residual-strength rate dependence and uncertainty.
5. Frictional timescales and the impact of climate change-driven extreme weather on rainfall-triggered landslides in Mizoram, NE India
Core Problem: Mizoram has very high landslide frequency, but physics-based timing models for rainfall-triggered slope failure are still scarce.
Key Innovation: Applies a rate-and-state friction block-slider framework to 19 landslides to relate pore-pressure transients, frictional timescales, and extreme rainfall.
6. MAF-UNet: A multimodal attention fusion U-Net for automatic loess sinkhole detection and mapping
Core Problem: Soil-piping sinkholes are small, diffuse, and hard to map from remote sensing, yet they threaten infrastructure and agricultural land.
Key Innovation: Introduces a multimodal attention-fusion U-Net that integrates spectral, texture, and terrain cues for automated loess sinkhole detection.
7. Hazard prediction of backfill and dumping zones of open-pit mining areas using InSAR and multi-factor ensemble modeling
Core Problem: Anthropogenic backfill and dumping zones combine consolidation settlement with potential slope failure, but deformation signals are difficult to separate from terrain and DEM errors.
Key Innovation: Builds an InSAR-based ensemble framework that accounts for consolidation and multi-factor controls to improve hazard prediction in open-pit mining landforms.
8. A coupled ICFD-DEM-FEM approach for modeling of debris flow impacts on flexible barriers
Core Problem: Flexible barrier design requires simultaneous representation of slurry flow, coarse particles, and nonlinear structural response.
Key Innovation: Couples incompressible CFD, DEM, and FEM to simulate debris-flow impacts on flexible barriers with explicit fluid-particle-structure interaction.
9. Soil erosion characteristics of terraces under extreme rainstorm conditions
Core Problem: Terraced landscapes can fail through gullies, caves, rills, slumping, and shallow landslides during extreme rainstorms, but controlling factors remain poorly separated.
Key Innovation: Uses UAV-derived high-resolution DEMs and field interpretation to relate terrace erosion forms to topographic and rainfall-driven controls on the Loess Plateau.
10. How rainfall reactivates ground fissures in loess regions: Deformation, stress response, and fissure propagation
Core Problem: Pre-existing ground fissures in loess terrains can reactivate during heavy rainfall, but the coupled deformation and stress pathway is not well constrained.
Key Innovation: Uses large-scale experiments and deformation-stress monitoring to clarify how rainfall drives fissure propagation and urban geo-hazard reactivation.
11. A UAV-Based Multi-Modal Vision System for Automated Sideslope Deformation Monitoring and Hazard Detection
Core Problem: Manual inspection of expressway slopes is inefficient and hazardous, especially where slow deformation precedes failure.
Key Innovation: Develops a UAV-based multimodal vision workflow for automated large-area slope observation, deformation monitoring, and hazard detection.
12. Seismic Stability Assessment of Rock Slopes at Tunnel Portals Using Integrated Numerical and Response Surface Methods
Core Problem: Tunnel portals in jointed rock are vulnerable to planar and wedge failures under seismic loading, but empirical ratings alone cannot resolve scenario uncertainty.
Key Innovation: Integrates kinematic analysis, slope-mass classification, FEM-DFN modeling, pseudo-static loading, and response surfaces for portal-slope stability assessment.
13. An integrated reconstruction framework of flash flood to unravel impacts, controlling factors and disaster amplification
Core Problem: Post-event flash-flood analysis often simulates hydrology without reconstructing impacts, controls, and amplification pathways in one framework.
Key Innovation: Combines hydrological-hydrodynamic reconstruction with impact and causal analysis to diagnose how flash floods evolve and intensify in mountain settings.
14. A data-driven framework for predicting downstream flood extent from Dez Dam releases using machine learning algorithms and remote sensing
Core Problem: Reservoir releases can produce downstream flood impacts that are hard to translate rapidly into spatial exposure estimates.
Key Innovation: Combines Landsat observations, dam-release records, and machine-learning models to map inundation extent and exposed assets downstream of Dez Dam.
15. Evaluation and attribution of flood disaster loss risk based on Bayesian network structure learning: a case study of the urban agglomeration in the middle Yangtze River
Core Problem: Flood-loss risk depends on interacting hazard, exposure, and vulnerability factors whose causal structure is difficult to infer from conventional indices.
Key Innovation: Uses Bayesian network structure learning to evaluate flood-loss risk and attribute controlling factors across the middle Yangtze River urban agglomeration.
16. Probabilistic seismic hazard assessment for Taiwan: Updates and improvements in TEM PSHA2025
Core Problem: National seismic hazard models must incorporate changing source, ground-motion, and regional evidence while maintaining consistency for engineering use.
Key Innovation: Updates the Taiwan Earthquake Model PSHA with expanded source coverage, international event integration, and methodological improvements for TEM PSHA2025.
17. Convective and non-convective nature of Balearic meteotsunamis: a 50-year historical review
Core Problem: Meteotsunami risk in the Balearic Islands depends on atmospheric disturbance type, but long-term event classification remains incomplete.
Key Innovation: Analyzes 191 events over 50 years to distinguish convective and non-convective pressure disturbances behind Ciutadella meteotsunamis.
18. Investigating the occurrence of an unusual tidal event along the north-west coast of Ireland
Core Problem: Unusual coastal water-level events can be misclassified as storm surge, seiche, freak wave, meteotsunami, or tsunami without integrated forensic analysis.
Key Innovation: Compares multiple event mechanisms for a February 2022 Irish coastal event using observational timing and meteorological-oceanographic context.
19. The characteristics of wind-blown snow hazards and an in-situ wind tunnel experiment study on snow disaster prevention along railway
Core Problem: Railway snow-drift mitigation is often based on indoor experiments that cannot reproduce field temperature, humidity, and snow-particle conditions.
Key Innovation: Introduces mobile outdoor wind-tunnel experiments and field monitoring to test snow-disaster prevention along railway lines.
20. Assessment and Mitigation of Cavity Induced Sinkhole Effects on Shallow Foundations Using FELA and Machine Learning
Core Problem: Subsurface cavities reduce shallow-foundation bearing capacity, yet the combined effects of cavity geometry, soil strength, and mitigation remain hard to generalize.
Key Innovation: Uses adaptive finite-element limit analysis and machine learning to evaluate and mitigate bearing-capacity loss above single and dual cavities.
21. Identification of soil erosion hotspot zones using integrated machine learning and geospatial models
Core Problem: Soil erosion hotspot mapping requires models that integrate topographic, climatic, vegetation, hydrological, and geological factors without relying on one index.
Key Innovation: Tests integrated machine-learning and geospatial models across 16 conditioning factors to delineate erosion-susceptibility zones.
22. Evaluating the effectiveness of the valley floor width parameter and comparing traditional and modern weighting methods in MCDM for flood hazard mapping
Core Problem: Semi-quantitative flood hazard maps remain sensitive to input geomorphic parameters and weighting schemes.
Key Innovation: Evaluates valley-floor width and compares entropy, Best-Worst, and modern MCDM weighting methods for flood-hazard mapping.
23. How much vulnerable the flood-affected villagers are? An assessment of livelihood vulnerability to flood hazard in Assam, India
Core Problem: Village-scale flood vulnerability is often generalized from coarse indicators, obscuring how household sensitivity and adaptive capacity vary within flood-prone districts.
Key Innovation: Applies the LVI-IPCC framework with 31 sub-components to assess livelihood vulnerability among flood-affected villages in Morigaon, Assam.
24. Coseismic Ionospheric Disturbances Detection With Multi-GNSS Case Study: Turkey Double Earthquakes on 6 February 2023
Core Problem: Large earthquakes perturb the ionosphere, but detecting and separating such signals requires dense GNSS evidence and careful timing.
Key Innovation: Uses multi-GNSS slant TEC observations to characterize ionospheric disturbances from the 2023 Turkey double earthquakes.
25. Bayesian three-dimensional seismic travel-time tomography for active- and passive-source seismic data using physics-informed neural network
Core Problem: Three-dimensional travel-time tomography is ill-posed, and conventional Bayesian gridded methods struggle with uncertainty quantification at scale.
Key Innovation: Uses a physics-informed neural network within a Bayesian framework to estimate 3D seismic velocity and uncertainty from active and passive data.
26. DETECT: seismological data from multiple dense arrays deployed across the Irpinia fault system in southern Italy
Core Problem: Fault-system hazard analysis depends on dense, well-archived seismic observations, but such datasets are often difficult to reuse.
Key Innovation: Publishes one year of continuous recordings from 200 temporary stations across the Irpinia fault system with standards-compliant metadata.
27. RAPID: A Reproducible Multi-Agent Pipeline for Interpretable Disaster Damage Assessment from Satellite and Street-View Imagery
Core Problem: Disaster damage assessment must generalize across data sources and scenes while remaining interpretable enough for operational review.
Key Innovation: Builds a reproducible multi-agent pipeline that combines satellite and street-view imagery for interpretable damage assessment under domain shift.
28. Comparison of Open Access Global DEMs for Geomorphological Applications in Coastal Areas Using LiDAR Data
Core Problem: Choosing DEMs by nominal resolution alone can mislead coastal geomorphological interpretation where vertical accuracy and geometric consistency matter.
Key Innovation: Compares open-access global DEMs against LiDAR to evaluate fitness for geomorphological applications in coastal terrain.
29. A Morpho-Phase Feature-Based Method for Geometric Error Mitigation in InSAR Image Matching
Core Problem: Geometric errors in InSAR image matching can degrade deformation measurement, especially in complex terrain and hazard-monitoring settings.
Key Innovation: Uses morpho-phase features to reduce geometric matching error and improve the reliability of InSAR-derived observations.
30. Dynamics of underground soil loss and its response to rainfall based on karst spring monitoring
Core Problem: Underground soil erosion in karst regions is difficult to quantify because losses occur through fissures and conduits rather than only surface channels.
Key Innovation: Uses karst spring monitoring as an outlet-scale measure to track underground soil-loss dynamics and rainfall response.
31. Research progress on the spatial distribution and monitoring of Benggang erosion
Core Problem: Benggang erosion is widespread in red-soil hilly regions, yet current maps and monitoring methods remain uneven and partly dependent on older surveys.
Key Innovation: Reviews spatial-distribution knowledge and monitoring approaches for Benggang erosion to identify gaps in regional erosion surveillance.
32. Comparative analysis of traditional and machine learning approaches for estimating the bearing capacity of granular soils
Core Problem: Foundation safety depends on reliable bearing-capacity estimates, but traditional formulas can vary substantially under simplified soil assumptions.
Key Innovation: Compares analytical bearing-capacity methods with machine-learning approaches to assess predictive performance for granular soils.
33. Mechanism of hydrophobic modification-induced reduction in matric suction on the inhibition of crack development in expansive soil under wet-dry cycles: Insights from molecular dynamics simulation
Core Problem: Wet-dry cycles drive cracking in expansive soils, but the microscale mechanism by which hydrophobic modification reduces matric suction is not fully resolved.
Key Innovation: Combines wet-dry cycle tests, suction measurements, contact-angle analysis, and molecular dynamics to explain crack inhibition.
34. Chemo-mechanical deterioration of oolitic limestone in simulated rhizosphere environments: Differential weathering and shielding mechanisms
Core Problem: Vegetation-based slope engineering changes the rhizosphere chemistry, but its effect on carbonate rock durability is uncertain.
Key Innovation: Tests organic and inorganic acid attack on oolitic limestone using mechanical and mineralogical observations to quantify differential deterioration.
35. Thermal fracture of granite under 9 GHz microwave irradiation: The critical effect of grain size
Core Problem: Microwave-assisted rock breakage depends on mineral heterogeneity, but grain-size effects on thermal fracture remain poorly constrained.
Key Innovation: Combines experiments and microstructural interpretation to show how granite grain size controls 9 GHz microwave-induced fracture behavior.
36. Experimental study on the effect of fracture structure on the mechanical properties of rocks in uniaxial compression
Core Problem: Rock-mass stability depends on fracture structure, yet uniaxial strength and failure mode vary with fracture geometry in ways that are hard to generalize.
Key Innovation: Uses controlled uniaxial compression experiments to isolate how fracture structure changes mechanical behavior and failure patterns.
37. Non-invasive hydro-geophysical imaging of landfill leachate source-to-plume dynamics using electromagnetic induction methods, geochemistry and hydrological modelling
Core Problem: Landfill leachate can migrate through heterogeneous subsurface deposits, but source zones and plume dynamics are difficult to resolve with single-method surveys.
Key Innovation: Integrates electromagnetic induction, geochemistry, and hydrological modelling to image leachate source-to-plume dynamics non-invasively.
38. Excavation-induced principal stress rotation and deformation responses in shallow-buried tunnels
Core Problem: Shallow-buried tunnels experience stress redistribution near the ground surface, but principal stress rotation is often simplified in design analysis.
Key Innovation: Uses three-dimensional finite-element modeling to compare staged excavation and end-state assumptions under stress-rotation effects.
39. Principal-stress rotation effects on multi-scale fracturing in deep rock tunnels with staggered structural planes: Physical modelling and DEM simulations
Core Problem: Deep tunnels with staggered structural planes can fracture across scales when excavation changes the direction of principal stresses.
Key Innovation: Combines physical modelling and DEM simulations to quantify stress-rotation effects on multi-scale fracturing in deep rock tunnels.
40. Remote sensing-based drought monitoring in northwestern Bangladesh: leveraging Google Earth Engine for comprehensive assessment
Core Problem: Agricultural drought in northwestern Bangladesh requires repeated, spatially consistent monitoring to support crop and water-management decisions.
Key Innovation: Uses Google Earth Engine to assess spatiotemporal drought variables from 2016 to 2024 across agricultural landscapes.
41. Brief communication: Multi-hazard-to-health-outcome (MH2O) pathways: the known, the unknown, and ten most urgent priorities
Core Problem: Climate-driven hazards often produce interacting health outcomes, but pathway knowledge remains fragmented across hazard types and sectors.
Key Innovation: Synthesizes known and unknown multi-hazard-to-health pathways and identifies ten urgent research priorities for compound risk assessment.
42. Per-Span Microwave Frequency Fiber Interferometry in Subsea Cables for Scalable Deep-Ocean Geophysical Monitoring
Core Problem: Deep-ocean geophysical monitoring is sparse and expensive, limiting observation of storms, tides, and teleseismic earthquakes.
Key Innovation: Demonstrates per-span microwave-frequency fiber interferometry on a 1,770 km operational subsea cable to resolve ocean and seismic signals.
43. VFM-CAKD: Category-Aware Knowledge Distillation From Visual Foundation Model for Aerial Imagery Semantic Segmentation
Core Problem: Lightweight aerial-image segmentation models are needed for drone deployment but often underuse the semantic knowledge in visual foundation models.
Key Innovation: Introduces category-aware knowledge distillation from a visual foundation model to improve efficient aerial imagery segmentation.
44. Evaluating and Enhancing Negation Comprehension in Remote Sensing MLLMs
Core Problem: Remote-sensing MLLMs must reason about absent or false objects, such as non-flooded evacuation routes, but negation understanding is under-tested.
Key Innovation: Introduces RS-Neg to evaluate negation comprehension from region to scene scale and proposes training strategies to improve remote-sensing MLLMs.
45. WaveUNet: An efficient deep network based on discrete wavelet transform and attention mechanism for InSAR phase unwrapping
Core Problem: Phase unwrapping remains a bottleneck for InSAR deformation products in steep terrain, dense fringes, and decorrelated hazard-monitoring scenes.
Key Innovation: Combines discrete wavelet transforms and attention in an efficient network for InSAR phase unwrapping under difficult terrain and noise conditions.
46. Exceedance Probabilities for Large Earthquakes from DIY Local Earthquake Ensemble Nowcasting and Forecasting: Magnitude, Natural Time, and Calendar Time
Core Problem: Operational earthquake outlooks require transparent probability estimates across magnitude, natural-time, and calendar-time representations.
Key Innovation: Compares local earthquake ensemble nowcasting and forecasting approaches for estimating exceedance probabilities of large earthquakes.
47. Leveraging Large Language Models for Agent-Based Simulation of Human-Water System Interactions
Core Problem: Human behavior strongly shapes hydrological extremes, but agent-based models often rely on simplified hand-coded decision rules.
Key Innovation: Explores large language models as natural-language reasoning agents for representing decisions in human-water system simulations.
48. Scene-Adaptive Camera Network Construction for Efficient Nap-of-the-Object Photogrammetry
Core Problem: Object-level UAV photogrammetry over large scenes is limited by inefficient camera-network design and frequent manual intervention.
Key Innovation: Builds scene-adaptive camera-network construction to improve close-range UAV photogrammetry for irregular terrain and infrastructure objects.
49. Autonomous Subsea Cable Search and Tracking with Graph-Optimised Priors and Visual Tracking
Core Problem: Subsea cables are vulnerable to natural hazards, but inspection is difficult when route maps are uncertain and cables are partly buried.
Key Innovation: Uses graph-optimized priors and visual tracking to support autonomous underwater search and tracking of subsea cable infrastructure.