TerraMosaic Daily Digest: Mar 31, 2026
Daily Summary
This March 31, 2026 digest distills 42 selected papers from 1694 analyzed records. The strongest papers mark an operational turn in slope-hazard science, replacing fixed rainfall proxies and static factor maps with time-varying, process-aware models. Kalman-filter observers reconstruct slow-moving landslide state and displacement, Transformer architectures ingest full rainfall sequences for multi-type landslide forecasting, and material point simulations resolve wetting-induced failure in collapsible loess. Across tower landslides, loess clusters, debris-avalanche retention, and seismic mountain slopes, the common lesson is that forecasting improves when the controlling state variable is made explicit.
A second strand broadens geohazard intelligence from slopes to infrastructure and recovery systems. Liquefaction damage models, rockburst warning thresholds, rockfall block characterization, avalanche forecast transfer tests, thaw-settlement risk, flood-road restoration sequencing, and multi-hazard screening frameworks all prioritize decision usefulness over stand-alone classification accuracy. Across the set, the strongest methods are interpretable, threshold-aware, and explicitly aimed at corridors, networks, and asset portfolios exposed to cascading failure.
Key Trends
Today’s strongest papers convert slope and geotechnical processes into time-dependent, threshold-aware, and infrastructure-facing decision frameworks.
- Rainfall and groundwater forcing are being modeled as trajectories rather than scalar triggers: the most informative landslide papers now use full rainfall sequences, evolving water tables, or state observers instead of fixed cumulative thresholds alone.
- Observation systems are moving from mapping toward quantification: segmentation models, InSAR, field monitoring, point clouds, and fiber optics are increasingly combined to estimate displacement state, mobilized volume, and instability geometry in operationally useful terms.
- Geotechnical mechanics are being translated into threshold-aware decisions: liquefaction damage, rockburst warning, wedge failure, and thaw settlement are increasingly expressed as warning criteria, risk classes, or probabilistic design quantities rather than isolated laboratory behaviors.
- Multi-hazard resilience is shifting from exposure description to recovery logic: shelter allocation, road reconnection, emergency-response behavior, and screening diagnostics are being treated as integral components of geohazard planning rather than downstream consequences.
Selected Papers
This digest features 42 selected papers from 1694 papers analyzed.
1. Reconstruction and forecasting of slow-moving landslide displacement using a Kalman Filter approach
Core Problem: Slow-moving landslides need better ways to reconstruct displacement histories and evolving soil properties from sparse observations while still supporting physically meaningful forecasts.
Key Innovation: The paper couples a simplified viscoplastic landslide model with a Kalman-filter-style observer, reset logic, and a new tuning strategy to jointly reconstruct displacement and unknown parameters and forecast motion from water-table inputs.
2. A comparative study on spatial patterns and dominant factors of landslide susceptibility under geomorphological differentiation
Core Problem: Large-scale landslide susceptibility mapping often misses local controls because spatial heterogeneity and geomorphological differentiation are not handled explicitly.
Key Innovation: The study compares susceptibility patterns and dominant controlling factors across geomorphologically differentiated terrain to improve region-specific landslide modelling.
3. Field investigation and numerical simulation of transmission tower landslides under rainfall
Core Problem: Critical transmission infrastructure on unstable slopes needs a mechanistic account of how rainfall infiltration and coupled loading drive deformation and failure.
Key Innovation: The study combines field investigation, laboratory testing, and transient numerical simulation to resolve hydro-mechanical and seismic controls on a tower-slope landslide and evaluate drainage and retaining countermeasures.
4. Transformer, more than meets the eye: A deep learning approach to integrate rainfall time-series in multi-type landslide probability modelling
Core Problem: Static rainfall proxies limit event-specific landslide probability models, especially when different landslide types respond to rainfall history in different ways.
Key Innovation: The paper feeds continuous rainfall time series into a Transformer network coupled with static predictors to model landslide-type-specific probabilities and interpret rainfall influence with SHAP-based gradients.
5. MPM analysis of wetting-induced failure in collapsible loess slopes
Core Problem: Standard numerical studies of loess slopes still struggle to capture wetting collapse, hydro-mechanical coupling, and post-failure behavior needed to explain real failures.
Key Innovation: The study develops an advanced material point framework that incorporates wetting-induced collapse, density-dependent hydraulic behavior, and strain softening, then validates it and back-analyzes a real loess landslide.
6. A combined semantic segmentation and time-series InSAR method for debris flow mobilized material volume evaluation
Core Problem: Debris-flow hazard management still lacks robust estimates of mobilized material volume when single mapping methods are used in isolation.
Key Innovation: The paper proposes a hybrid framework that combines semantic segmentation with time-series InSAR to evaluate debris-flow mobilized material volume for hazard prevention.
7. A comparative assessment of supervised models for landslide susceptibility mapping: a case study of Qiongzhong county, Hainan Island, China
Core Problem: Land-use planning in tropical landslide-prone terrain depends on knowing which supervised models actually perform best in a specific regional setting.
Key Innovation: The study compares six supervised algorithms for landslide susceptibility mapping in Qiongzhong County and shows ensemble-learning models, especially XGBoost and Random Forest, are strongest.
8. Exploring the spatial distribution patterns of loess landslides and identifying the key driving factors in typical regions of the upper Yellow River, China
Core Problem: Loess-landslide mitigation in the upper Yellow River region requires clearer evidence of where failures cluster and which environmental factors dominate their occurrence.
Key Innovation: The study maps loess-landslide susceptibility in Gaolan County and identifies distance to roads, annual rainfall, and elevation as the strongest drivers.
9. Liq-EBM: data-driven assessment of liquefaction triggering and associated ground damage
Core Problem: Liquefaction hazard evaluation needs a more accurate and transparent way to predict both triggering and settlement-related ground damage at engineering and regional scales.
Key Innovation: The paper introduces an explainable boosting machine workflow that predicts liquefaction triggering and maximum-shear-strain-based settlement from six standard inputs and validates it with case histories and a Taipei Basin application.
10. Seismic-induced dynamic response and instability characteristics of high mountain slopes in the Yarlung Tsangpo Grand Canyon, Tibetan Plateau, China
Core Problem: High mountain slopes in the Eastern Himalayan Syntaxis are earthquake-weakened, but their dynamic response and failure modes remain poorly constrained.
Key Innovation: The study combines field seismic recordings from two slopes with FLAC3D simulations to distinguish their dominant frequencies, response characteristics, and contrasting earthquake-triggered failure modes.
11. Green selective retention infrastructures for debris avalanche mitigation
Core Problem: Flow-like landslides need mitigation designs that can reduce runout and downstream impacts without simply shifting the hazard laterally.
Key Innovation: The study uses two-phase r.avaflow simulations at Cervinara to test selective retention barriers and quantify both their protective value and their side effects on flow routing.
12. Interpretable slope stability evaluation and optimization method based on hybrid extreme gradient boosting regression
Core Problem: Slope engineering needs predictive tools that can both explain stability drivers and optimize design parameters for safer configurations.
Key Innovation: The paper integrates a metaheuristic-optimized XGBoost regression model with SHAP interpretability and NSGA-II optimization to build a slope stability prediction-and-optimization workflow.
13. Critical angles of a rock bridge in slopes containing intermittent joints: Fracture mode identification and stability assessment
Core Problem: Intermittently jointed rock slopes are difficult to assess because rock-bridge fracture modes and their transition angles are not well defined.
Key Innovation: The paper uses numerical experiments and theory to identify fracture-mode angle ranges, derive a critical transition angle, and embed that result in a stability model.
14. Hydro-meteorological and Infrastructural Damage Analysis of the Recent Ramban Cloudburst Event in the North-Western Himalayan Region of Jammu and Kashmir, India
Core Problem: Mountain transport corridors need integrated event reconstruction to understand how extreme precipitation cascades into slope failure, flooding, and infrastructure damage.
Key Innovation: The paper fuses WRF simulations, multi-temporal InSAR, Bayesian damage mapping, and SWOT hydrology to reconstruct the April 2025 Ramban cloudburst and its cascading impacts.
15. Geospatial, geophysical & geotechnical investigation of glacial depositions in Sikkim Himalayas
Core Problem: Hazard planning in debris-covered Himalayan glacier terrain needs integrated evidence on deposit properties, supraglacial change, and moraine stability.
Key Innovation: The paper combines satellite indicators, MASW, GPR, laboratory geotechnics, microstructural analysis, and slope-stability assessment to identify highly failure-prone glacial deposits in Sikkim.
16. Geotechnical insights from on-site investigations of the 2023 Pazarcık-Elbistan Earthquake Sequence in Kahramanmaraş, Türkiye
Core Problem: The 2023 Kahramanmaraş earthquake sequence requires field-based analysis to explain how local soil conditions amplified shaking and produced liquefaction, lateral spreading, and cyclic softening damage.
Key Innovation: The study combines strong-motion records, microzonation boreholes, and field observations to link basin and near-fault effects with observed geotechnical failures and code-level implications.
17. Determination of critical stress for rockburst warning: A model incorporating borehole-sensor interaction and its validation
Core Problem: Rockburst monitoring systems still lack a physically grounded way to set warning thresholds instead of relying on empirical stress alarms.
Key Innovation: The paper derives and validates a borehole-sensor interaction model that computes critical warning stress from rock strength, brittleness, mining stress, and borehole geometry.
18. An analytical framework to assess static versus dynamic triggering of fault-slip rockbursts
Core Problem: Fault-slip rockburst studies often blur the triggering roles of static and dynamic stress changes around tunnels.
Key Innovation: The paper introduces an analytical framework that separates static, dynamic, and dual triggering, builds a hazard map around a fault, and demonstrates the method on a historical tunnel event.
19. From the Swiss Alps to the Pyrenees: Evaluating the transferability of machine learning models for avalanche forecasting
Core Problem: Operational avalanche centers need to know whether machine-learning forecasting models trained in one snow climate remain useful when transferred to another mountain region.
Key Innovation: The paper tests Swiss operational avalanche models in the Pyrenees and shows that they retain moderate predictive skill, supporting cautious cross-region deployment after validation.
20. Integrating 3D Point Cloud analysis for potentially unstable rock blocks characterization: a method for assessing size and shape distribution
Core Problem: Rockfall risk analysis needs more reliable characterization of unstable block size and shape than simplified field methods typically provide.
Key Innovation: The study integrates point-cloud segmentation, 3D meshing, OMBB-derived flattening, and uncertainty analysis to generate robust block-size distributions and improved shape classification.
21. Probabilistic analysis of thaw settlement and serviceability in Arctic embankments: A case study on the ITH
Core Problem: Arctic infrastructure managers need a probabilistic framework to quantify thaw-settlement uncertainty and serviceability loss under permafrost degradation.
Key Innovation: The study links TEMP/W with Monte Carlo analysis and maintenance optimization to estimate thaw-settlement risk and serviceability classes along the Inuvik-Tuktoyaktuk Highway.
22. Integrated risk-based design approach for wedge hazard in underground mining drifts
Core Problem: Underground drift design needs a realistic way to estimate where structurally controlled wedges will form and what risk they impose.
Key Innovation: The paper combines DFN modelling, surveyed excavation geometry, and economic metrics into a workflow for locating critical wedges and optimizing reinforcement decisions.
23. Mapping global post-earthquake ecosystem damage boundaries
Core Problem: Earthquake damage studies still lack a consistent way to delineate ecosystem-scale damage boundaries beyond mapping individual physical failures.
Key Innovation: The study builds an automated global framework that uses Landsat change detection, kernel density, thresholding, and morphology to map post-earthquake ecosystem damage boundaries for 284 earthquakes.
24. Influence of fines content and initial static shear on the cyclic liquefaction behavior of silty sands: A DEM study
Core Problem: Engineers still lack a unified mechanistic explanation for how low fines content and initial static shear jointly control cyclic liquefaction in silty sands.
Key Innovation: Using DEM triaxial simulations, the paper shows that the initial state parameter better organizes undrained behavior and clarifies how fines and anisotropic consolidation alter liquefaction resistance.
25. A coupled B-C-H-W model for predicting wave-induced residual liquefaction of bio-grouted seabed around buried tunnels
Core Problem: Offshore tunnel design needs a way to predict wave-driven residual liquefaction while quantifying how bio-grouting changes seabed response.
Key Innovation: The paper develops a coupled biological-chemical-hydraulic-wave model that simulates MICP treatment and demonstrates strong liquefaction suppression under optimized grouting layouts.
26. Experiments and elastoplastic analyses on soil disturbance of soft clay subjected to cyclic loading
Core Problem: Seismic design practice often underestimates how cyclic loading disturbs structured soft clays and reduces post-cyclic undrained shear strength.
Key Innovation: The study combines cyclic triaxial tests with a structure-aware elastoplastic model to show that structural degradation, not remolding alone, drives the observed strength loss.
27. A time-varying critical acceleration framework for an embedded cantilever retaining wall in saturated sand
Core Problem: Conventional seismic displacement models for retaining walls do not capture transient pore-pressure softening and unloading recovery in saturated backfill.
Key Innovation: The study introduces a time-varying critical-acceleration Newmark framework, calibrated with centrifuge data, that updates resistance through excess-pore-pressure softening and recovery indicators.
28. A strain-dependent modified pseudo-dynamic analysis of expanded landfill stability under leachate conditions
Core Problem: Seismic stability analyses of expanded landfills can misestimate factor of safety when they ignore strain dependence and leachate-driven pore-pressure effects.
Key Innovation: The paper develops a modified pseudo-dynamic framework that incorporates equivalent-linear strain dependence, waste heterogeneity, and leachate buildup to identify critical loading phases.
29. Failure behavior and energy evolution of sandstone under coupled compression-shear loading and partial seawater immersion: Effects of immersion height and inclination angle
Core Problem: Subsea mining pillars face poorly constrained strength loss and failure-mode changes when compression-shear loading interacts with long-term partial seawater immersion.
Key Innovation: The paper quantifies energy evolution and shows that seawater-driven microstructural deterioration shifts sandstone toward weaker, more ductile mixed tensile-shear failure.
30. Deep-learning full-waveform inversion of snowpack GPR: joint permittivity-resistivity imaging for snow-soil hydrological mapping
Core Problem: Snowpack GPR inversion remains too computationally expensive and initialization-sensitive for operational mapping of snow and underlying soil properties.
Key Innovation: The paper proposes a deep full-waveform inversion framework that jointly recovers permittivity and resistivity from GPR and produces coherent snow-water and soil-moisture maps.
31. Revealing the dynamics and multidimensional resilience of rainstorm-flood cascade disasters in mountain valley cities: An interpretable machine learning case study from Southwestern China
Core Problem: Mountain-valley cities need clearer evidence for how rainstorm-flood cascades evolve and how resilience shifts across multiple urban dimensions.
Key Innovation: The study uses interpretable machine learning to diagnose cascade-disaster dynamics and multidimensional resilience patterns in a Southwestern China case study.
32. A framework of evidence-based multi-hazard risk screening for disaster risk reduction and resilience planning
Core Problem: Decision-makers still lack a rapid and evidence-based way to screen large portfolios of assets for multi-hazard risk before committing to detailed analysis.
Key Innovation: The paper develops a bottom-up screening framework that combines exposure, vulnerability, criticality, coping capacity, and hotspot ranking for multi-hazard DRR planning.
33. A Graph-Based Deep Learning Approach for Daily Flash Flood Susceptibility Modeling in China
Core Problem: Flash-flood susceptibility forecasting still struggles to represent daily variability and upstream-downstream dependence across large catchment networks.
Key Innovation: The paper uses graph-based deep learning to build a daily flash-flood susceptibility model that explicitly leverages hydrologic connectivity across China.
34. Urban road connectivity assessment and impassable-road restoration sequencing under the impact of flood events
Core Problem: Flood-disrupted road systems need methods that move beyond static vulnerability to restoration order and recovery prioritization.
Key Innovation: The study evaluates connectivity loss and derives restoration sequences for impassable roads under flood scenarios, directly linking hazard impact to recovery logic.
35. Analysis of the structural characteristics and deformation mechanisms of unstable rock masses in the red beds along the Renchi Expressway
Core Problem: High-positioned red-bed unstable rock masses along transport corridors are difficult to characterize quantitatively because cavity growth, weathering, and structural control interact across scales.
Key Innovation: The paper combines UAV photogrammetry and particle-flow simulation to resolve cavity-controlled deformation, weathering-driven strength loss, and failure mechanisms along the Renchi Expressway.
36. Formation mechanisms and modes of karst collapse columns based on multi-information exploration technology: a case study of Fengpei coalfield, North China
Core Problem: Karst collapse columns remain difficult to characterize mechanistically despite their direct role in mine water inrush and collapse risk.
Key Innovation: The study uses integrated exploration data to classify collapse-column formation modes and link them to low-disturbance treatment implications in the Fengpei coalfield.
37. The influencing factors of emergency response hesitancy in extreme precipitation events in Northern China
Core Problem: Warning systems lose effectiveness when exposed populations hesitate to act during extreme precipitation events.
Key Innovation: The paper identifies the trust, comprehension, cost, and behavioral factors that delay emergency response under extreme precipitation, informing more actionable warning design.
38. Enhancing random field-based limit equilibrium analysis of road embankment failures on soft soils using convolutional neural networks
Core Problem: Probabilistic embankment-stability analysis under spatially variable soft soils remains computationally expensive for rapid failure screening.
Key Innovation: The paper uses convolutional neural networks to accelerate random-field-based limit-equilibrium analysis for road embankment failure on soft soils.
39. Monitoring 3D movement of structures and soil masses using fiber optic cables under the notion of finite element method
Core Problem: Distributed sensing workflows still struggle to reconstruct full three-dimensional soil and structure movement in geotechnical settings.
Key Innovation: The paper extends fiber-optic sensing into a finite-element-informed workflow for reconstructing 3D movement of soil masses and structures.
40. Water flow timing, quantity, and sources in a fractured high mountain permafrost rock wall
Core Problem: Permafrost rock-wall stability remains difficult to assess because fracture-water timing, quantity, and origin are rarely monitored directly in high mountains.
Key Innovation: The paper provides direct observations of fracture water flow and source contributions in a high mountain permafrost rock wall, linking hydrology to instability assessment.
41. Collapse behavior of thick interlayers in bedded salt caverns during the leaching phase
Core Problem: Salt-cavern creation still lacks a clear account of how thick interlayers fail and collapse during the leaching stage.
Key Innovation: The paper identifies and validates collapse modes that govern underground cavity instability during salt-cavern development.
42. Erosion of sandy soils and pavement subsidence induced by drainage pipe leakage: insights from macroscopic experiments and microscopic simulations
Core Problem: Urban leakage-driven internal erosion remains poorly resolved despite its direct progression toward void growth and pavement subsidence.
Key Innovation: The paper combines macroscopic experiments and microscopic simulations to track the evolution from pipe leakage to sand erosion, void growth, and surface subsidence.