Initiated by Dr. Xin Wei, University of Michigan
Ongoing development by the community

TerraMosaic Daily Digest: Mar 17, 2026

March 17, 2026
TerraMosaic Daily Digest

Daily Summary

The March 17, 2026 literature shows a decisive move toward cascade-aware hazard analysis. The strongest studies resolve how destructive outcomes emerge not from isolated triggers, but from transitions between stages: moraine landslides become GLOFs through impulse-wave transfer and breach initiation; supraglacial-lake drainage becomes debris-flow disaster through valley confinement and entrainment; rainfall-driven slopes fail through evolving pore-pressure and clay-migration states; and flood timing, post-fire runoff, and coastal surge risk become predictable only when lag structure and regional transfer are modeled explicitly. The scientific emphasis is shifting from event labels toward process chains.

A parallel shift is methodological. Many of today's best papers improve hazard intelligence by encoding physical or spatial structure directly into the workflow: watershed connectivity in flash-flood mapping, sample representativeness in landslide susceptibility, topographic amplification in tunnel fragility, height-aware lidar projection in permafrost thaw mapping, and constitutive realism in liquefaction or slope-reliability analysis. Engineering studies on debris-flow baffles, drainage abrasion, roots, geogrids, piles, scour, and erosion-protection materials show the same pattern. Mitigation is becoming process-specific, with design choices tied to the exact instability mechanism they are intended to control.

Key Trends

Today's strongest papers connect initiation, amplification, and control, making hazard assessment more mechanistic and more operational at the same time.

  • Hazard cascades are being modeled end to end: landslide entry, wave generation, breach growth, debris entrainment, flood onset, and downstream exposure are increasingly treated as one connected system.
  • Susceptibility mapping is becoming structure-aware: watershed topology, geographical similarity, event-specific conditioning, and sample-ratio control now materially change predictive reliability.
  • Hidden hydraulic and material states are central to failure explanation: pore-water pressure, soil suction, cyclic softening, scour evolution, and cryosphere preconditioning are being used to explain when hazards actually intensify.
  • Mitigation studies are sharply failure-mode specific: baffles, roots, drainage structures, geogrids, anti-slide piles, and erosion-protection materials are evaluated by the exact instability they suppress.
  • Operational warning frameworks are recovering lead time from process structure: paired storms, seasonal lag coupling, small-sample surrogates, and corrected hazard simulations all extend usable forecasting under sparse data.

Selected Papers

This digest features 35 selected papers from 1336 papers analyzed across landslide-GLOF cascades, flood timing and coastal surge risk, permafrost and glacier degradation, susceptibility and vulnerability mapping, and process-specific geotechnical mitigation.

1. Dynamic analysis of moraine landslide-induced glacial lake outburst flood event

Source: Landslides Type: Hazard Modelling Geohazard Type: Landslide-triggered GLOF Relevance: 10/10

Core Problem: The full dynamic chain from moraine landslide impact to breach initiation and downstream GLOF amplification remains weakly quantified in alpine basins.

Key Innovation: Survey-constrained numerical reconstruction shows how impulse-wave transfer, terminal-dam failure, and elevation-drop-driven momentum gain transformed a local slope failure into a destructive flood cascade.

2. Remote sensing interpretation of causes and dynamics of the Gyirong outburst flood disaster, Tibet, on July 8, 2025

Source: Natural Hazards Type: Detection and Monitoring Geohazard Type: GLOF-debris-flow cascade Relevance: 10/10

Core Problem: Supraglacial-lake failures are difficult to interpret because downstream geomorphic transitions can drastically amplify discharge and transform hazard type.

Key Innovation: Remote sensing and sedimentological analysis reveal that the 2025 Gyirong disaster intensified about sixfold after entering narrow, sediment-rich V-shaped valleys, where the GLOF evolved into a debris flow.

3. Coseismic landslide susceptibility modelling via machine learning associated with the September 5, 2022, Luding Mw 6.8 earthquake, southeastern Tibetan Plateau

Source: Bull. Eng. Geol. & Env. Type: Susceptibility Assessment Geohazard Type: Earthquake-induced landslides Relevance: 9/10

Core Problem: The spatial controls on Luding earthquake landslides remain difficult to rank across topographic, hydrologic, and seismic gradients.

Key Innovation: More than 1000 machine-learning runs show that topography, road proximity, river proximity, and fault distance dominate the coseismic pattern while low-complexity models remain robust.

4. From snowmelt to disaster: modeling debris flow dynamics in the Shatoobar Nala watershed, Hunza, Pakistan

Source: Natural Hazards Type: Risk Assessment Geohazard Type: Debris flows Relevance: 9/10

Core Problem: Snowmelt-triggered debris flows in high mountain valleys are hard to manage when initiation, runout, and infrastructure exposure are studied separately.

Key Innovation: RAMMS-DF back-analysis links the 2022 event to prolonged temperature-driven snowmelt, then converts calibrated dynamics into scenario-based hazard, vulnerability, and risk maps.

5. Robustness assessment of motorway networks under rainfall-induced landslides

Source: Natural Hazards Type: Risk Assessment Geohazard Type: Rainfall-induced landslides Relevance: 9/10

Core Problem: Road systems in mountain terrain need network-scale evaluation of landslide disruption, not just site-by-site slope diagnosis.

Key Innovation: The study fuses susceptibility mapping, Monte Carlo landslide scenarios, and graph restoration metrics to identify locally critical motorway links and faster recovery strategies.

6. Landslide susceptibility assessment based on Bayesian optimization and geographical similarity

Source: Natural Hazards Type: Susceptibility Assessment Geohazard Type: Landslides Relevance: 9/10

Core Problem: Susceptibility maps are degraded by uncertain sample ratios and weakly representative non-landslide samples.

Key Innovation: Bayesian optimization of the positive-to-negative ratio, combined with geographical-similarity stratified sampling, improves RF and CNN mapping accuracy and yields more credible zoning.

7. Flood Risk Follows Valleys, Not Grids: Graph Neural Networks for Flash Flood Susceptibility Mapping in Himachal Pradesh with Conformal Uncertainty Quantification

Source: ArXiv (Geo/RS/AI) Type: Susceptibility Assessment Geohazard Type: Flash floods Relevance: 8/10

Core Problem: Pixel-based flash-flood maps ignore watershed connectivity, even though upstream routing is a primary control on downstream hazard.

Key Innovation: A GraphSAGE model built on sub-watershed connectivity outperforms grid baselines and adds conformal uncertainty bounds, showing that valley structure itself is a predictive asset.

8. Effects of Pore Water Pressure on the Stability of Granular Slopes Under Simulated Rainfall

Source: Geotech. & Geol. Eng. Type: Concepts & Mechanisms Geohazard Type: Rainfall-induced slope failure Relevance: 8/10

Core Problem: The combined effect of intense rainfall and clay content on granular-slope collapse remains difficult to resolve experimentally.

Key Innovation: Multi-sensor physical experiments identify a three-stage failure sequence in which pore-pressure rise, runoff drag, and clay migration jointly accelerate settlement and enlarge landslide volume.

9. Morphological and dynamic characteristics of abrasion on drainage structures in response to debris flow

Source: J. Mountain Science Type: Mitigation Geohazard Type: Debris-flow impact on drainage structures Relevance: 8/10

Core Problem: Damage to drainage channels and check dams under repeated debris-flow abrasion lacks a clear process-based design framework.

Key Innovation: Field investigations show that flow velocity, watershed cutting degree, and fine-particle content control abrasion intensity, enabling more targeted structure design and repair.

10. Systematic Evaluation and Correction of Extreme Water Level in Global Storm Surge Simulation

Source: GRL Type: Hazard Modelling Geohazard Type: Coastal flooding and storm surge Relevance: 8/10

Core Problem: Global storm-surge models show large systematic errors in gauge-sparse regions, weakening their value for coastal hazard assessment.

Key Innovation: A transferable quantile-mapping framework calibrated at tide gauges improves extreme-water-level estimation well beyond observed sites and extends useful correction into sparse coastlines.

11. Paired Storms Approach Reveals Post-Fire Flood Characteristics and Drivers

Source: Water Resources Research Type: Hazard Modelling Geohazard Type: Post-fire floods Relevance: 8/10

Core Problem: Post-fire flood forecasting is limited by sparse event records and the difficulty of comparing disturbed and undisturbed storm responses.

Key Innovation: A paired-storm framework matches post-fire floods to pre-fire analogues and shows that most post-fire peak flows are at least twice their pre-fire counterparts.

12. Seasonal and upstream rainfall controls on monsoon flood onset in Bangladesh's Haor wetlands

Source: Natural Hazards Type: Early Warning Geohazard Type: Monsoon floods Relevance: 7/10

Core Problem: Flood onset in the Haor wetlands is difficult to anticipate because inundation depends on both local and upstream rainfall with delayed coupling.

Key Innovation: Multi-decadal satellite inundation and rainfall analysis identifies a robust 2-3 month rainfall-to-flood lead time and a clearly lengthened flood season after 2001.

13. Research on the application of the extreme learning machine model based on the slime mold optimization algorithm in landslide tsunami disaster early warning under small-sample tasks

Source: Natural Hazards Type: Early Warning Geohazard Type: Landslide-generated tsunamis Relevance: 7/10

Core Problem: Rare landslide-tsunami events rarely provide enough training data for high-capacity warning models.

Key Innovation: An SMA-ELM surrogate outperforms CNN, MLP, and several standard baselines under small-sample conditions, improving end-to-end tsunami prediction for warning applications.

14. The role of El Niño Southern Oscillation in driving coastal hazards in the U.S. Pacific Northwest

Source: NHESS Type: Hazard Modelling Geohazard Type: Coastal flooding and erosion Relevance: 7/10

Core Problem: ENSO has long been assumed to control Pacific Northwest coastal hazard episodes, but the teleconnection remains poorly constrained.

Key Innovation: A stochastic climate emulator shows that the ENSO-coastal-hazard relationship is weaker and more conditional than commonly assumed, refining how teleconnections should inform risk management.

15. Spatiotemporal patterns and zonation of typhoon and non-typhoon extreme rainfall hazards in the typical coastal region of southeastern China

Source: Geomatics, Nat. Haz. & Risk Type: Hazard Modelling Geohazard Type: Extreme rainfall hazards Relevance: 7/10

Core Problem: Extreme-rainfall hazard assessments in coastal East Asia often fail to distinguish typhoon-driven from non-typhoon rainfall regimes.

Key Innovation: The study separates and zones the spatiotemporal behavior of typhoon and non-typhoon rainfall extremes, improving regional hazard characterization in southeastern China.

16. An Improved LightGBM model with ADASYN and Whale Optimization Algorithm for Rockburst Intelligent Prediction

Source: Rock Mech. & Rock Eng. Type: Early Warning Geohazard Type: Rockbursts Relevance: 7/10

Core Problem: Rockburst prediction suffers from class imbalance and unstable performance in data-driven warning models.

Key Innovation: ADASYN balancing combined with WOA-LightGBM improves prediction accuracy across benchmark and engineering datasets, strengthening intelligent rockburst risk assessment.

17. The role of roots in riverbank stabilization under extreme fluctuating water conditions: From the perspective of riparian plant diversity

Source: Catena Type: Mitigation Geohazard Type: Riverbank erosion and bank-slope instability Relevance: 7/10

Core Problem: Vegetation-based bank protection is widely advocated, but the stabilizing value of plant diversity under repeated inundation remains poorly quantified.

Key Innovation: Root and vegetation surveys, combined with infinite-slope modelling, show that different diversity indices regulate shallow and deep failure differently under fluctuating water levels.

18. Dynamic behavior and shock morphology of rapid granular flows impacting a finite-width baffle within an inclined channel

Source: Engineering Geology Type: Mitigation Geohazard Type: Rapid granular mass flows Relevance: 7/10

Core Problem: The impact behavior of rapid granular flows against barrier structures remains insufficiently quantified for robust mitigation design.

Key Innovation: Experiments derive phase relations and scaling laws for shock morphology around finite-width baffles, clarifying how inflow regime, roughness, and geometry control protective performance.

19. Preserving Vertical Structure in 3D-to-2D Projection for Permafrost Thaw Mapping

Source: ArXiv (Geo/RS/AI) Type: Detection and Monitoring Geohazard Type: Permafrost thaw Relevance: 6/10

Core Problem: Standard 3D-to-2D aggregation of lidar features destroys the vertical forest structure needed to infer permafrost thaw depth.

Key Innovation: A height-aware projection decoder with stratified sampling preserves canopy-to-ground structure and improves UAV-based thaw-depth mapping in boreal terrain.

20. A winter-season meteotsunami in the Balearics: the 22 January 2021 squall-line driven event

Source: Natural Hazards Type: Early Warning Geohazard Type: Meteotsunamis Relevance: 6/10

Core Problem: Winter meteotsunamis in the western Mediterranean remain poorly characterized and difficult to forecast because most documented events are warm-season cases.

Key Innovation: Dense observations and prediction-system analysis document an unusual squall-line-driven winter meteotsunami, improving process understanding and predictability for this rare coastal hazard.

21. Photogrammetric 3D reconstruction and 3D printing of a Moso Bamboo root system for geotechnical testing and soil reinforcement analysis

Source: Bull. Eng. Geol. & Env. Type: Mitigation Geohazard Type: Vegetated slope reinforcement Relevance: 6/10

Core Problem: Natural root systems are too variable and fragile for controlled geotechnical testing of bio-reinforced slope behavior.

Key Innovation: Photogrammetric reconstruction plus 3D printing creates mechanically realistic bamboo-root replicas that enable reproducible soil-reinforcement testing.

22. Appraisal of the HySand_base constitutive model against the Karlsruhe fine sand database

Source: Acta Geotechnica Type: Hazard Modelling Geohazard Type: Soil liquefaction Relevance: 6/10

Core Problem: Reliable constitutive simulation of cyclic sand behavior is essential for liquefaction assessment, yet model performance varies across loading conditions.

Key Innovation: Broad validation against the Karlsruhe fine sand database shows that HySand_base reproduces pore-pressure buildup and cycles to liquefaction across diverse test regimes.

23. Effect of Topographic Amplification on Seismic Vulnerability of Mountain Tunnels Using Multi-stripe Analysis

Source: Rock Mech. & Rock Eng. Type: Vulnerability Geohazard Type: Seismic vulnerability of mountain tunnels Relevance: 6/10

Core Problem: Tunnel seismic fragility assessments often ignore how hill geometry and tunnel position change amplification and damage probability.

Key Innovation: Multi-stripe fragility analysis shows that topographic amplification can increase extensive-damage probability by 18%-375%, arguing for location-specific seismic design.

24. Experimental and Numerical Study of Pull-Out Behaviour and Stability of Geogrid-Reinforced High Embankments

Source: Geotech. & Geol. Eng. Type: Mitigation Geohazard Type: Reinforced embankment instability Relevance: 6/10

Core Problem: High embankment stability depends on geogrid-soil interaction under realistic field conditions, but those couplings remain incompletely constrained.

Key Innovation: Pull-out tests and PLAXIS modelling show that moisture control dominates stability gains, while reverse-wrapping layouts significantly improve interfacial efficiency and safety factor.

25. Ground response analysis and liquefaction assessment of subsea soils at an offshore platform site

Source: Bull. Earthquake Eng. Type: Hazard Modelling Geohazard Type: Liquefaction and cyclic softening Relevance: 6/10

Core Problem: Offshore design needs better discrimination between classical liquefaction and cyclic softening in submarine soils under different earthquake return periods.

Key Innovation: Site-specific geotechnical data and nonlinear response analysis show that shallow cyclic softening, rather than widespread liquefaction, may control the governing seismic hazard.

26. Estimation of the initial shear modulus of unsaturated soil from soil-water characteristic curve

Source: Engineering Geology Type: Concepts & Mechanisms Geohazard Type: Unsaturated-soil slope failure Relevance: 6/10

Core Problem: Small-strain stiffness in unsaturated soils strongly affects collapse and slope response, yet current predictive relations remain weakly physical.

Key Innovation: A spring-based model derives initial shear modulus from soil-water characteristic behavior by linking stiffness to the connectivity of dry and wet fractions.

27. Fracture evolution and strata failure mechanisms in ultra-thick coal seams during staged mining: Insights from physical and numerical modeling

Source: Engineering Geology Type: Concepts & Mechanisms Geohazard Type: Mining-induced ground failure Relevance: 6/10

Core Problem: Ultra-thick coal extraction produces complex staged overburden failure that remains difficult to resolve across physical and numerical scales.

Key Innovation: Coupled physical modelling, UDEC simulation, DIC, and fractal analysis reveal a hierarchical two-stage failure sequence with evolving fracture connectivity and subsidence structure.

28. Assessing vulnerability of rural buildings to tornadoes and their relationships with building attributes and surrounding land uses

Source: IJDRR Type: Vulnerability Geohazard Type: Tornado impacts Relevance: 6/10

Core Problem: Rural tornado vulnerability remains poorly quantified where building properties and surrounding land use interact to shape damage intensity.

Key Innovation: Logistic regression and XGBoost show how both building attributes and village-scale land-use context jointly control tornado damage patterns in rural Jiangsu.

29. Efficacy and mechanisms of hydrosedimentary regulation in a typical loess plateau watershed under different governance stages

Source: Catena Type: Mitigation Geohazard Type: Watershed erosion and check-dam regulation Relevance: 6/10

Core Problem: It remains unclear how much older check-dam systems can still reduce sediment hazards as storage capacity declines.

Key Innovation: A distributed model reveals a regime shift from transport limitation to supply limitation, indicating that stronger hillslope source control can partly compensate for aging gully sinks.

30. Temperature sensitivity and rainfall heat flux drive rapid mass loss of low-latitude glaciers in the Southeastern Qinghai-Tibet Plateau

Source: Catena Type: Concepts & Mechanisms Geohazard Type: Glacier retreat and GLOF precursors Relevance: 6/10

Core Problem: The joint contribution of warming and rain heat flux to low-latitude glacier retreat remains insufficiently resolved.

Key Innovation: Surface-energy-balance modelling shows that rainfall heat flux contributes materially to temperate-glacier ablation and helps explain unusually rapid retreat in the southeastern QTP.

31. Subdomain sampling method for efficient reliability analysis of pile-reinforced slopes involving copula-based cross-correlated random fields

Source: Computers and Geotechnics Type: Mitigation Geohazard Type: Pile-reinforced slope instability Relevance: 6/10

Core Problem: Reinforced-slope reliability is hard to quantify realistically when soil strength varies spatially and non-Gaussian dependence is important.

Key Innovation: The proposed subdomain sampling method preserves copula-based variability while cutting safety-factor evaluations by about 90%, and clarifies when anti-slide piles change failure mode.

32. Encroachment of riparian vegetation in a braided river due to catchment management in Japan

Source: Catena Type: Mitigation Geohazard Type: River instability and hydrological hazards Relevance: 5/10

Core Problem: The cumulative effect of dams, channel works, and sediment trapping on braided-river hazard behavior remains difficult to disentangle.

Key Innovation: Long-term analysis shows how multiple catchment interventions stabilized gravel bars, enabled woody encroachment, and altered channel form in ways that matter for hydrological hazard management.

33. Experimental study on the bearing performance of high pile cap foundation under the coupling effect of scouring and horizontal cyclic loads

Source: Soil Dyn. & Earthquake Eng. Type: Vulnerability Geohazard Type: Scour-affected offshore foundations Relevance: 5/10

Core Problem: Offshore foundation safety under combined scour and cyclic loading remains poorly constrained because erosion is often idealized as static cover loss.

Key Innovation: Coupled tank experiments track the joint evolution of scour morphology and cyclic response, showing how cyclic loading accelerates pit development and degrades bearing capacity.

34. Reuse of improved shield muck as reinforcement material for bridge erosion protection

Source: Transportation Geotechnics Type: Mitigation Geohazard Type: Bridge erosion protection Relevance: 5/10

Core Problem: Bridge erosion protection needs low-cost, field-validated materials that can resist underwater scour while remaining constructible.

Key Innovation: Laboratory and field tests show that improved shield muck can meet fluidity, shear-stress, permeability, and durability requirements for practical bridge erosion protection.

35. Viscoelasticity and yield behavior of loess mudflows: experimental insights under varied density and temperature conditions

Source: Frontiers in Earth Science Type: Concepts & Mechanisms Geohazard Type: Loess mudflows Relevance: 7/10

Core Problem: Loess-flow initiation and mobility are hard to predict without controlled measurements of how density and temperature shift rheological thresholds.

Key Innovation: Rotational-rheometer experiments quantify how density and temperature alter yield stress, consistency, and viscoelastic response, improving the physical basis for modelling loess-flow initiation and propagation.