TerraMosaic Daily Digest: May 15, 2026
Daily Summary
May 15's papers are led by unusually coherent work on failure transition. The Blatten landslide study isolates accelerating sliding-related seismicity before collapse, while the Mt Job reconstruction shows how a small volcanic rock avalanche became a long-runout supraglacial debris flow through snow, ice, water, and altered-rock interactions. Two other core landslide papers turn difficult internal and downstream states into measurable quantities: active landslide thickness inferred from SAR interferometric phase, and breach hydraulics moderated by geobag chains during landslide-dam emergency drainage.
The same state-variable logic extends across permafrost, faults, volcanoes, and hydrologic hazards. Permafrost-slope experiments separate thermal disturbance from warm-humid compound forcing; granular-flow and debris-flow papers resolve mobility through fragmentation, rheology, and velocity traces; and fault-volcano studies show how normal-stress perturbations, rifting, and pulsed sulfur emissions shape unstable slip, unrest, and terrestrial impacts. The broader set strengthens the forcing and observation layer through rainfall-sensing datasets, diffusion nowcasting, soil-moisture SAR, subcanopy DEM correction, reservoir-operation modelling, drought-propagation diagnostics, and infrastructure recovery metrics.
Key Trends
The day is organized around transition diagnostics: precursor seismicity, active thickness, breach erosion, thaw-front retreat, granular fragmentation, fault stress timing, and hydrologic forcing uncertainty.
- Landslide research is moving from detection to state inference: seismic precursors, multiphase runout, SAR-derived active thickness, tree-ring movement records, and breach-control experiments estimate initiation, propagation, and downstream release.
- Cold-region slope hazards are coupled thermal-hydrological-mechanical systems: permafrost slope tests, warm-humid thaw slumping, subsea permafrost inversion, frozen-sediment permeability, and glacier basal-state seismology quantify phase change and weakening.
- Fault, volcanic, and seismic studies emphasize timing: normal-stress perturbations, Reykjanes unrest, volcanic sulfur pulses, site-response learning, and offshore foundation fragility convert dynamic forcing into thresholds or impact pathways.
- Hydrologic forcing is being densified: opportunistic rainfall sensing, diffusion nowcasting, sub-daily rainfall disaggregation, reservoir operation, soil-moisture SAR, and drought-propagation diagnostics improve hazard inputs.
- Infrastructure and erosion papers link degradation to service risk: water inrush, tunnel fracture evolution, embankment erosion, expansive-soil stiffness, CO2 storage digital twins, equitable recovery, and inclusive volcanic warnings tie material state to consequence.
Selected Papers
This issue contains 50 selected papers from 1,300 papers analyzed. The leading papers target failure transition directly: Blatten seismic precursors, Mt Job rock-avalanche-to-debris-flow mobility, SAR-derived landslide active thickness, landslide-dam breach control, permafrost slope retrogression, thaw slumping under warm-humid forcing, granular-flow fragmentation, debris-flow velocity traces, fault stress perturbations, and Reykjanes volcano-tectonic coupling. The broader set extends the same state-estimation logic to subsea permafrost, tsunami mitigation, glacier basal conditions, volcanic sulfur-pulse impacts, DInSAR coregistration, CO2-storage digital twins, mine-floor water inrush, physics-guided seismic response, rainfall sensing, diffusion precipitation nowcasting, SAR soil moisture, drought propagation, tunnel stability, erosion mechanics, and inclusive volcanic early warning.
1. Seismic Precursors to the Blatten, Switzerland Landslide Revealed by Unsupervised Machine Learning
Core Problem: The transition from stable slope motion to catastrophic collapse is rarely observed clearly enough to separate rockfall noise from sliding-related precursors.
Key Innovation: Continuous seismic records from the 2025 Blatten failure are clustered with unsupervised machine learning, revealing accelerating sliding-related signals during the final two days before collapse.
2. Characterisation and multiphase flow modelling of a rock avalanche turned into a supraglacial debris flow: Insights from Mt Job, British Columbia, Canada
Core Problem: A small volcanic rock avalanche at Mt Job travelled far beyond its initial volume, requiring explanation of how snow, ice, water, and altered rock amplified mobility.
Key Innovation: Photogrammetry, remote sensing, field instrumentation, conceptual process analysis, and multiphase r.avaflow simulations reconstruct the 2023 Mt Job rock-avalanche-to-debris-flow cascade.
3. Inferring landslide active thickness from SAR-derived interferometric phases
Core Problem: Slow-moving landslide hazard assessment needs active thickness estimates, but thickness is usually inferred indirectly from sparse drilling or simplified geometry.
Key Innovation: A SAR interferometric-phase method is introduced to infer landslide active thickness directly from deformation signals, linking satellite phase to internal landslide structure.
4. Experimental study on flood control during landslide dam breach using geobag chains
Core Problem: Emergency spillways can lower landslide-dam reservoirs but may also intensify erosion, lateral instability, and downstream peak discharge.
Key Innovation: Physical experiments test geobag chains as an erosion-control measure during landslide-dam breach drainage, quantifying how installation changes breach erosion and flood release.
5. Retrogressive failure mechanisms of ice-rich permafrost slopes under thermal disturbance on the Qinghai–Tibet plateau
Core Problem: Ice-rich permafrost slopes are destabilizing under warming, yet the mechanics of ground-ice melt, retrogressive collapse, and slope deformation remain underconstrained.
Key Innovation: Remote sensing, field investigation, and large-scale physical modelling are integrated to resolve failure mechanisms of Qinghai-Tibet Plateau permafrost slopes under thermal disturbance.
6. Warm-humid synergy aggravates thermal thaw slumping on permafrost slopes
Core Problem: Permafrost slope instability may respond differently to warming alone and to combined warming-humidification, but this compound forcing is rarely isolated experimentally.
Key Innovation: Model experiments simulate baseline, warming, and warm-humid scenarios to show how compound climate conditions deepen thermal disturbance and aggravate thaw slumping.
7. Dynamic particle fragmentation modulates friction evolution and shear mobility in dry granular flows
Core Problem: Catastrophic dry granular flows depend on shear localization and particle breakage, but micro-to-macro controls on frictional mobility are often treated qualitatively.
Key Innovation: Three-dimensional shear-zone tracking and fragmentation analysis show how dynamic particle breakage modulates friction evolution and mobility in landslide and rock-avalanche analogues.
8. Mudline interpretation for superelevation-based velocity estimation: A perspective for practical applications
Core Problem: Debris-flow hazard analysis often uses bend superelevation, but practical velocity estimates depend on how post-event mudlines are interpreted.
Key Innovation: The study formalizes mudline interpretation for superelevation-based velocity estimation, improving field translation from debris-flow traces to impact-relevant flow speed.
9. Do Normal Stress Vibrations Stabilize or Destabilize Faults? Experimental Perspectives
Core Problem: Small normal-stress changes from natural forcing or fluid injection can either stabilize or destabilize faults, but the controlling timescale is poorly constrained.
Key Innovation: Hydrothermal carbonate-gouge experiments show that perturbation period relative to instability recurrence controls whether stress vibration reinforces stick-slip or promotes quieter slow slip.
10. Dynamic Coupling Between Faulting, Rifting and Magmatism During 2021‐2025 Unrest on Reykjanes Peninsula, Iceland
Core Problem: The interaction of diking, surface rupture, earthquakes, and magma release is rarely captured through a multi-year volcanic unrest sequence.
Key Innovation: High-resolution seismicity, focal mechanisms, satellite geodesy, surface deformation, and eruption records reveal how 2021-2025 Reykjanes unrest partitioned extension among rifting, faulting, and magmatism.
11. Mapping Subsea Permafrost Distribution in the Canadian Beaufort Sea With Marine Seismic and Deep Learning
Core Problem: Subsea permafrost degradation may reshape Arctic shelf geohazards and carbon transport, but its depth-resolved distribution is difficult to map regionally.
Key Innovation: A multi-input, multi-output neural network estimates velocity and attenuation along Beaufort Sea seismic lines, resolving spatial variability and thinning of upper subsea permafrost near the shelf edge.
12. Reassessing seawall height and building setback distance for tsunami mitigation: A DualSPHysics-based hydrodynamic study
Core Problem: Tsunami mitigation is often framed around seawall height alone, while setback distance and impact timing jointly control building forces.
Key Innovation: DualSPHysics 3D hydrodynamic simulations quantify how seawall geometry and building setback alter tsunami force attenuation and arrival timing.
13. Temporal Variations in Seismic Ambient Noise Identify Seasonal Changes in Basal Conditions at Sermeq Kujalleq in Kangia, Greenland
Core Problem: Subglacial water pressure and basal decoupling are central to glacier dynamics but are hard to observe directly near fast-flowing ice streams.
Key Innovation: Continuous seismic ambient-noise and GNSS data at Sermeq Kujalleq identify the melt-season transition from strong ice-bed coupling to partial basal decoupling.
14. Possible use of tree-growth disturbances to record landslide movements as a complement to conventional monitoring techniques and remotely sensed data analysis in forested areas
Core Problem: Forested landslides often lack long monitoring records because field instrumentation is sparse and remote sensing may miss subcanopy deformation.
Key Innovation: Tree-growth disturbances are evaluated as a complementary archive for reconstructing landslide movement histories and reactivation triggers in forested terrain.
15. Cyclic instability of gap-graded soils with suffusion-induced fines loss and initial static shear: A DEM study
Core Problem: Natural slopes can lose fines during heavy rainfall and then fail under cyclic loading, but the coupled micromechanics of suffusion and static shear are difficult to observe.
Key Innovation: DEM simulations show how fines loss reduces liquefaction resistance, shifts failure mode toward flow failure, and accelerates instability under initial static shear.
16. Optimisation-based calibration of advanced constitutive models for debris-flow deposited mixed-grain soils
Core Problem: Debris-flow deposits are difficult to parameterize for engineering models because broad grain-size distributions make testing costly and sparse.
Key Innovation: An optimization framework calibrates advanced constitutive models for debris-flow-deposited mixed-grain soils using minimal triaxial test data and dynamic response matching.
17. Fracture evolution and sliding-to-fracturing transition in layered rock under direct shear: Insights from integrated AE-DIC-ultrasonic monitoring
Core Problem: Layered rock masses in slopes, tunnels, and foundations can fail progressively, but the transition from sliding to fracturing is hard to capture at multiple scales.
Key Innovation: Integrated acoustic-emission, DIC, and ultrasonic monitoring resolves fracture evolution and anisotropic shear failure in bedded sandstone.
18. An Improved Coregistration Strategy Integrating Local Iterative Pixel Offset Tracking and Weighted ESD for Sentinel-1 TOPS Mode DInSAR Measurement in Nonstationary Scenarios
Core Problem: Large nonstationary azimuth deformation can corrupt Sentinel-1 TOPS coregistration and propagate phase jumps into deformation products.
Key Innovation: A local iterative pixel-offset tracking and weighted ESD strategy improves DInSAR coregistration under nonstationary tectonic or ground-deformation scenarios.
19. Toward Real-Time Digital Twins for CO₂ Storage: Graph Neural Network Ensembles for Multi-physics Forecasting and Risk Quantification
Core Problem: Secure CO2 storage needs rapid updates of plume migration, geochemical alteration, and surface displacement without running full THMC simulators for every scenario.
Key Innovation: Graph neural network ensembles emulate multi-physics CO2 storage simulations and quantify uncertainty in gas saturation, pH, and vertical surface displacement.
20. Study on hydraulic properties in a pore-fracture coupled model for predicting water inrush from mine floors
Core Problem: Mining-induced failure zones create coupled pore-fracture pathways that control water inrush from coal-seam floors.
Key Innovation: A pore-fracture coupled model simulates complete-rock and fracture conductivity to evaluate how fracture structure governs water pressure and flow beneath mine floors.
21. Physics-guided deep learning for nonlinear site response under earthquakes using KiK-net data: An LSTM case study
Core Problem: Earthquake ground-motion prediction needs nonlinear site response estimates that generalize beyond purely data-driven station learning.
Key Innovation: A physics-guided LSTM embeds nonlinear-to-linear response constraints into the loss function, improving accuracy, generalization, and interpretability for seismic site response.
22. Mapping Localized Permafrost and Seasonal Freezing with Machine Learning
Core Problem: Rapidly thawing localized permafrost poses infrastructure and planning risks but is difficult to map at municipal scale.
Key Innovation: Ground thermal monitoring, land-cover classification, and machine learning generate permafrost and seasonal-freezing maps for McGrath, Alaska.
23. OpenRainER: an open-source dataset for studying the opportunistic sensing of rainfall in Emilia-Romagna, Italy
Core Problem: Dense rainfall observations are needed for flash-flood and landslide-triggering rainfall analysis, but gauges and radars are unevenly distributed.
Key Innovation: OpenRainER releases one-minute commercial microwave-link data and processing tools for opportunistic rainfall sensing in Emilia-Romagna.
24. Short-Term Precipitation Forecast Based on Diffusion Spatiotemporal Network
Core Problem: Radar nowcasting often smooths intense precipitation structures that matter for urban flooding and slope-failure triggers.
Key Innovation: A ViT-modulated diffusion spatiotemporal network refines radar-echo forecasts and improves heavy-precipitation skill metrics on the HKO-7 benchmark.
25. An Integrated Remote-Sensing Framework for Channel Dynamics Monitoring in Braided Rivers
Core Problem: Braided rivers can shift rapidly, complicating floodplain exposure, sediment-routing assessment, and channel-management decisions.
Key Innovation: A remote-sensing framework combines surface-water extraction, boundary reconstruction, and river-geometry metrics to monitor 1986-2025 Lower Yellow River channel dynamics.
26. Linking Plant Traits to Fire Potential Mapping: A Feasibility Study in Australian Ecosystems
Core Problem: Fire-potential maps need physically meaningful flammability information rather than generic vegetation classes alone.
Key Innovation: Laboratory spectroscopy, plant traits, combustion experiments, and satellite scaling are linked to map ecosystem-scale fire potential in Australia.
27. Dynamic response and seismic fragility of monopile foundations for offshore wind turbines considering pile-soil interaction
Core Problem: Offshore wind foundations in seismic regions need fragility estimates that account for soil stiffness, pile geometry, and structural frequency shifts.
Key Innovation: Finite-element validation and fragility analysis quantify how pile-soil interaction changes monopile response and seismic vulnerability.
28. Pulsed volcanic sulfur emissions linked to the end-Triassic terrestrial crisis
Core Problem: Volcanic crises can damage terrestrial systems through pulsed sulfur emissions and acid-rain pathways, but the timing of atmospheric forcing and ecosystem response is hard to constrain.
Key Innovation: Science Advances links mercury, sulfur isotopes, sulfur-bearing molecular fossils, biomarkers, and wildfire indicators to pulsed CAMP volcanic sulfur deposition during the end-Triassic terrestrial crisis.
29. Integrating Social Equity into Resilience-based Post-disaster Recovery of Interdependent Infrastructure Systems
Core Problem: Infrastructure recovery models often optimize speed while leaving unequal service restoration across communities unmeasured.
Key Innovation: A multi-objective framework uses the Gini coefficient to quantify horizontal equity in interdependent infrastructure recovery after the Wenchuan earthquake.
30. T-ACD: A topography-aware change detection method for mapping high-resolution soil moisture with time-series SAR
Core Problem: Antecedent soil moisture is central to flood and landslide triggering, but SAR retrievals degrade over complex topography and vegetation.
Key Innovation: A topography-aware change-detection method explicitly models local-incidence-angle effects to improve time-series SAR soil-moisture mapping.
31. Measure Less, Map More: Using Machine Learning, Physiography, and Prior Depth Maps to Extrapolate In‐Swath Snow Depth Measurements Across Mountain Basins
Core Problem: Mountain snowpack controls runoff, flood timing, and slope loading, yet basin-wide snow-depth surveys remain expensive and sparse.
Key Innovation: Swath-random forests use narrow airborne or drone snow-depth swaths, physiography, and prior maps to extrapolate basin-wide snow depth.
32. A Hybrid Empirical and Conceptual Model for Improving the Representation of Reservoirs With Limited Data in Hydrological Models
Core Problem: Large-scale hydrological models often simplify reservoirs despite their strong control on floods, droughts, and downstream water availability.
Key Innovation: A hybrid empirical-conceptual reservoir operation model reduces data needs while improving compatibility with basin hydrological simulations.
33. Gas Permeability of Frozen Sediments: Implications for Modeling Gas and Water Transport in Permafrost
Core Problem: Permafrost thaw and gas migration require permeability functions across frozen and unfrozen states, but direct frozen-sediment measurements are limited.
Key Innovation: Laboratory gas-permeability measurements validate modelling relationships for frozen sediment gas and water transport across ice saturations.
34. An Interpretable Physics‐Guided Deep Learning (IPGDL) Framework for Blue‐Green Water Robust Projection and Nonlinear Interpretation in the River Basin
Core Problem: Hydrological projection needs nonlinear learning without losing physically interpretable water-partitioning mechanisms.
Key Innovation: An interpretable physics-guided deep-learning framework couples SWAT mechanisms, CNN-BiLSTM learning, CMIP6 forcing, land-use scenarios, and SHAP interpretation.
35. A Semiempirical Modeling for Soil Moisture Retrieval Using High-Resolution SAR Data
Core Problem: Fine-scale SAR soil-moisture retrieval is limited by vegetation, roughness, and scale-dependent backscatter effects.
Key Innovation: A semiempirical SMAP-AVS model is evaluated with ALOS-2 PALSAR to improve high-resolution L-band soil-moisture retrieval.
36. Understanding meteorological, runoff, and agricultural drought propagation and their influencing factors in an ensemble of multiple datasets
Core Problem: Drought preparedness requires understanding how precipitation deficits propagate into runoff and soil-moisture impacts across regions.
Key Innovation: An ensemble analysis of ERA5, GLDAS, and TerraClimate quantifies drought response time, propagation rate, lag time, and controls using SHAP attribution.
37. The dynamics of spatio-temporal droughts in Northeast Brazil
Core Problem: Drought monitoring often lacks interpretable metrics for how drought areas grow, persist, and retreat through time.
Key Innovation: A three-dimensional drought analysis introduces growth, state, and dynamic curves to classify recurrent drought-evolution patterns.
38. A KED-based framework for enhanced subcanopy topography retrieval leveraging TanDEM-X InSAR and spaceborne LiDAR data
Core Problem: Forested terrain remains a major source of DEM error for landslide, flood, and erosion modelling.
Key Innovation: A KED-based framework decomposes InSAR DEM residuals and integrates spaceborne LiDAR control points to improve subcanopy topography retrieval.
39. Terrace failure and soil loss induced by seepage in a coarse-grained brown soil
Core Problem: Terraces reduce erosion but can fail when seepage pathways weaken coarse-grained soils on sloping farmland.
Key Innovation: Constant-head seepage experiments quantify how land slope and bulk density control terrace failure and soil loss.
40. 7Be tracer observation of physical crust wind erosion in winter and spring: Influencing pathways and mechanisms of soil wind erosion
Core Problem: Wind erosion on sloping drylands depends on crust stability, slope position, and disturbance, but these pathways are hard to quantify.
Key Innovation: Beryllium-7 tracer observations and runoff-plot experiments identify how tillage disturbance and slope gradient regulate physical-crust wind erosion.
41. A modified CSLE for slope-scale soil loss prediction under different vegetation types in China
Core Problem: Slope-scale erosion estimates can misrepresent vegetation protection when understory and vegetation-type effects are simplified.
Key Innovation: Modified CSLE biological-control equations incorporate vegetation coverage and type to improve national-scale slope soil-loss prediction.
42. Deep learning-based adaptive multi-resolution characterization of CO2 storage sites from monitoring data
Core Problem: CO2 plume migration and storage-site properties are hard to infer from sparse monitoring data in heterogeneous reservoirs.
Key Innovation: An adaptive multi-resolution deep-learning characterization method updates subsurface-property representations from monitoring data.
43. Deep learning error post-processing improves stochastic watershed modeling
Core Problem: Flood and drought simulations need uncertainty-aware watershed models whose residuals remain valid under changing hydrologic states.
Key Innovation: Deep-learning post-processing reduces conditional dependence in stochastic watershed models and improves ensemble streamflow simulation.
44. Simulating sub-daily rainfall time series in the absence of sub-daily observations
Core Problem: Flood and landslide design often needs sub-daily rainfall, but many basins only have daily observations.
Key Innovation: A parsimonious framework combines daily rainfall generation with multifractal disaggregation using information from intensity-duration-frequency curves.
45. A novel multi-link approach to drought propagation analysis using modified dynamic time warping
Core Problem: Drought propagation metrics often miss time-varying links between meteorological, agricultural, and hydrological drought.
Key Innovation: A modified dynamic-time-warping approach captures flexible multi-link drought propagation across 179 catchments.
46. Is the Weibull distribution reliable to describe the rainfall kinetic power and momentum in Mediterranean environments?
Core Problem: Soil-erosion prediction depends on rainfall drop-size distributions, kinetic power, and momentum, but distributional assumptions need testing.
Key Innovation: More than 47,000 drop-size distributions are used to evaluate Weibull-based estimates of rainfall kinetic power and momentum.
47. Stability analysis and fracture evolution in topographical bias tunnels within layered rock masses: insights from FDEM simulation
Core Problem: Bias tunnels in layered rock masses can experience asymmetric stress redistribution and progressive fracture evolution after excavation.
Key Innovation: Improved FDEM simulations calibrated against physical tests quantify bedding and mechanical controls on tunnel deformation and fracture evolution.
48. A nonlinear unified hardening model for breakable siliceous mudstone coarse-grained soil under low confining pressure considering penetrating erosion
Core Problem: Embankments built with breakable siliceous mudstone soils can degrade mechanically as penetrating erosion and particle crushing progress.
Key Innovation: A nonlinear unified hardening model links erosion intensity, particle breakage, and critical-state shifts under low confining pressure.
49. Impact of seasonal hydrological changes on the small-strain shear modulus of undisturbed and remolded highly expansive soils
Core Problem: Highly expansive soils weaken under wetting-drying-freeze-thaw cycles, affecting slopes, foundations, and transport earthworks.
Key Innovation: Resonant-column tests quantify how seasonal hydrological changes alter small-strain shear modulus in undisturbed and remolded expansive soils.
50. The inclusivity of volcanic hazard early warning systems: Experiences of persons with disabilities in Indonesia
Core Problem: Volcanic early-warning systems can fail if warnings are inaccessible or not actionable for disabled residents near active volcanoes.
Key Innovation: A participatory mixed-methods study near Merapi evaluates warning access, coverage, message clarity, and social-network pathways for persons with disabilities.