TerraMosaic Daily Digest: Mar 30, 2026
Daily Summary
This March 30, 2026 digest distills 49 selected papers from 2011 analyzed records. The strongest papers advance a more process-explicit landslide science, in which failure is treated as the outcome of coupled hydrology, deformation, entrainment, and infrastructure interaction rather than as a static susceptibility state. Runoff-seepage coupling, reservoir-bank deformation networks, graph-neural surrogates for runout, debris-flow clogging models, and cold-region settlement studies all resolve how material state evolves into transfer, loading, and secondary hazard.
A second strand turns regional analysis into operational intelligence. Susceptibility studies now combine climate and seismic forcing, source-area constraints, road vulnerability, InSAR-derived deformation, and robust validation strategy; monitoring papers fuse InSAR, GNSS, groundwater, soil moisture, and electromagnetic perturbations to isolate mechanism rather than merely detect motion; and mitigation work extends from sustainable slope treatment and rockburst energy relief to village-scale and emergency-system resilience. Across the set, the most valuable contributions are those that connect process realism to decisions.
Key Trends
Today’s literature treats instability as a coupled trajectory, linking conditioning, triggering, transfer, infrastructure loading, and resilience rather than isolating any one stage.
- Coupled hydro-mechanical and post-failure models are replacing single-stage slope analysis: the leading papers explicitly connect runoff, preferential seepage, pore-pressure buildup, structural weakness, runout, and secondary loading rather than separating initiation from consequence.
- Susceptibility mapping is becoming dynamic, infrastructure-aware, and uncertainty-aware: climate-seismic forcing, source-area constraints, road vulnerability, deformation products, and repeated spatial validation are moving from optional context to core model ingredients.
- Monitoring is shifting from motion detection to mechanism diagnosis: multi-sensor frameworks increasingly use deformation, hydrology, and precursor signals to distinguish reservoir forcing, seasonal response, and pre-failure change in physically interpretable ways.
- Resilience is becoming a first-class geohazard outcome: mitigation materials, emergency-system design, village-scale vulnerability, and corridor-scale exposure are now being studied as integral parts of slope-hazard science rather than downstream add-ons.
Selected Papers
This digest features 49 selected papers from 2011 papers analyzed.
1. Cross-domain coseismic landslide segmentation: local boundary enhancement & global pixel contrastive learning
Core Problem: Rapid coseismic landslide mapping remains unreliable across domains when blurred scar boundaries and weak feature contrast degrade automated segmentation.
Key Innovation: The paper combines local boundary enhancement with global pixel-level contrastive learning to sharpen scar delineation and improve cross-domain landslide segmentation transfer.
2. Accelerating stochastic simulation of post-failure landslide runout using a random graph neural network-based simulator
Core Problem: Probabilistic post-failure landslide simulation is still too computationally expensive for robust hazard assessment under spatially variable soil properties.
Key Innovation: The study introduces a random graph neural network surrogate that emulates MPM-style runout behavior and enables large Monte Carlo analyses at much lower cost.
3. Insights into the deformation mechanisms of reservoir landslides: multidimensional observations and UDEC numerical simulations related to the Tanjiahe landslide (China)
Core Problem: Large ancient reservoir-bank landslides remain difficult to interpret when deformation, groundwater, and subsurface movement are monitored in isolation.
Key Innovation: The paper integrates InSAR, GNSS, inclinometer, and groundwater observations with UDEC simulations to resolve deformation controls and failure behavior under extreme conditions.
4. Hydrological system and stability of loess-mudstone interface slopes under rainfall: a bidirectionally coupled model of surface runoff and subsurface seepage
Core Problem: Loess-mudstone interface failures are hard to predict because runoff, preferential seepage, pore-pressure growth, and crack-tip stress interact in a strongly coupled way.
Key Innovation: The study develops and validates a bidirectionally coupled runoff-seepage-stability model that identifies dominant hydrologic pathways and failure modes under different rainfall regimes.
5. Efficient multi-source deep learning for rapid landslide mapping in the Karst mountains of Bijie, China
Core Problem: Rapid landslide mapping in karst terrain is hampered by optical look-alikes such as bare limestone and quarry benches that confuse automated detectors.
Key Innovation: The paper builds a multi-source deep-learning workflow that fuses complementary inputs to improve fast landslide delineation in complex karst mountains.
6. MCB3: Modeling clogging by boulders on barriers and bridges - A tool to simulate jamming caused by debris flows through series of cascading structures
Core Problem: Debris-flow hazard analysis still struggles to represent how boulder jamming at multiple bridges and barriers cascades through a channel network and alters downstream loading.
Key Innovation: The study extends a stochastic hydraulic jamming model so debris-flow hydrographs can be routed through consecutive structures to quantify cumulative clogging and protection effects under uncertainty.
7. Ensemble deep learning framework for landslide susceptibility mapping and road vulnerability index development in the Chittagong Hill Tracts, Bangladesh
Core Problem: Hazard studies in the Chittagong Hill Tracts rarely connect landslide-prone terrain to the way road infrastructure is actually exposed and disrupted.
Key Innovation: The paper couples ensemble deep-learning susceptibility mapping with a road vulnerability index so slope hazard patterns can be translated into corridor-level infrastructure risk.
8. Analysis of the causes of giant landslide clustering in the upper reaches of the Yellow River since the Holocene
Core Problem: Why giant landslides in the upper Yellow River cluster in space and time through the Holocene remains poorly resolved beyond broad tectonic and climatic explanations.
Key Innovation: The study combines inventory evidence, paleoclimate context, and structural setting to infer coupled dry-to-wet climate transitions and tectonic forcing behind giant-landslide clustering.
9. Integrating machine learning and deep learning for enhanced landslide susceptibility and propagation dynamics in Northwestern Pakistan
Core Problem: Reliable landslide susceptibility and propagation assessment remains difficult in steep mountainous terrain where conventional models miss interacting topographic and geological controls.
Key Innovation: The paper couples machine-learning and deep-learning susceptibility models with Flow-R runout simulation and PS-InSAR cross-checking, with a hybrid LSTM-XGBoost workflow performing best.
10. Elucidating loessal landslide initiation in wood- and shrub-land by hydro-mechanical heterogeneity
Core Problem: The hydrological and mechanical controls that distinguish woodland from shrubland initiation of loessal landslides remain poorly constrained.
Key Innovation: The study combines field surveys, soil-moisture monitoring, dye tracing, and root-mechanics testing to show how vegetation-dependent preferential flow and reinforcement alter failure behavior.
11. Landslide susceptibility mapping using ensemble learning integrating climate and seismic factors along the CPEC
Core Problem: Corridor-scale landslide susceptibility mapping still underrepresents the combined influence of climatic and seismic conditioning on transport-risk hotspots.
Key Innovation: The paper applies an ensemble-learning susceptibility framework that explicitly integrates climate and seismic factors along the China-Pakistan Economic Corridor.
12. Reducing subjectivity in landslide susceptibility mapping method: an integrated methodology
Core Problem: Heuristic landslide susceptibility methods remain sensitive to analyst judgment, limiting reproducibility and confidence in zonation results.
Key Innovation: The study integrates the SSEP framework with GIS-based bivariate statistics to build a more objective and testable susceptibility-mapping workflow.
13. Satellite-derived seasonal fluctuations in surface displacement and soil moisture: Implications for landslide activity
Core Problem: Seasonal displacement signals on slow-moving landslides are hard to interpret when hydrological drivers are poorly monitored in mountain terrain.
Key Innovation: The paper combines SBAS InSAR deformation time series with satellite-derived soil moisture to separate seasonal hydrologic motion from potentially hazardous landslide acceleration.
14. Integrating geospatial data and hybrid machine learning models for landslide susceptibility assessment in semi-arid environments
Core Problem: Accurate landslide susceptibility mapping in semi-arid terrain remains difficult because standard predictor sets and single-model workflows often transfer poorly.
Key Innovation: The study uses integrated geospatial predictors and hybrid machine-learning models to improve susceptibility assessment in semi-arid environments.
15. The multitemporal landslide inventory map of the Collazzone study area, central Italy
Core Problem: Open multitemporal landslide inventories remain rare despite their central value for hazard evaluation, trigger analysis, and landscape-process reconstruction.
Key Innovation: The paper releases a rigorously mapped multidecadal inventory of more than 3,500 landslides built from historical aerial photography, stereo satellite imagery, field checks, and an open layered dataset.
16. A KG-driven framework for enhanced post-failure landslide stability assessment
Core Problem: Post-failure landslide assessment during emergency response still relies heavily on incomplete data and expert judgment.
Key Innovation: The paper proposes a knowledge-graph-driven framework to organize fragmented post-failure information and support faster stability assessment.
17. Exploring the effect of soil zoning in the TRIGRS and Scoops3D integrated model on the stability of rainfall-induced shallow landslides
Core Problem: Physically based shallow-landslide modelling remains sensitive to how soil zones are represented within coupled infiltration and stability simulations.
Key Innovation: The study tests soil-zoning schemes inside an integrated TRIGRS-Scoops3D workflow to clarify their effect on rainfall-induced shallow-landslide stability analysis.
18. Analysis of the failure mechanisms and treatment effectiveness of the Houlang village landslide
Core Problem: Rainfall and excavation-triggered landslides in fault-controlled terrain are hard to stabilize when structural weakness, groundwater rise, and human disturbance act together.
Key Innovation: The paper combines field investigation, numerical stability analysis, and monitoring to diagnose the Houlang Village failure and verify a multi-measure stabilization system.
19. Formation mechanism of the Ninglang paleo-landslide at the northern segment of the Chenghai Fault Zone, Southeastern Tibetan Plateau, China
Core Problem: The triggering, long-term persistence, and breach history of the Ninglang paleo-landslide dam have remained incompletely constrained.
Key Innovation: Field investigation, UAV photogrammetry, and dating show a seismically triggered landslide dam whose eventual breach was governed by fault-related weakness and piping erosion.
20. Modeling the cumulative impact of river curvature on landslide-induced surge waves: refinement based on the Pan Jiazheng model
Core Problem: The widely used Pan Jiazheng relation loses accuracy in curved river reaches where curvature strongly alters surge-wave propagation after landslides.
Key Innovation: The study quantifies river curvature, introduces a curvature-aware computation length, and validates an improved surge-wave relation with FLOW-3D simulations and field cases.
21. Validation and prediction challenges in data-driven landslide susceptibility mapping: insights from random sampling of training and test data in machine learning models
Core Problem: Random train-test splitting can bias performance estimates and hide uncertainty in data-driven landslide susceptibility mapping.
Key Innovation: The paper benchmarks nine machine-learning methods across repeated resampling, multiple validation metrics, and statistical tests while exposing the performance loss under spatial cross-validation.
22. Evaluation of highway collapse hazard susceptibility based on the coupling of information value model and multiple machine learning models
Core Problem: Highway collapse hazards remain difficult to predict with single-model susceptibility workflows that underresolve the spatial controls threatening transport corridors.
Key Innovation: The study couples an information value model with multiple machine-learning models to improve highway collapse susceptibility prediction.
23. Refined collapse susceptibility assessment in Tonghua city based on collapse source area data and multi-model coupling
Core Problem: Steep terrain, rainfall, and human disturbance create urban-peripheral collapse hazards that are not well resolved by coarse susceptibility methods.
Key Innovation: The paper refines collapse susceptibility assessment by incorporating collapse source-area data into a coupled multi-model framework for Tonghua City.
24. Evaluation of spatial susceptibility to collapse based on random forest: a case study in the mountainous area of Changping District, Beijing
Core Problem: Peri-urban mountainous regions require better collapse susceptibility maps because engineering disturbance and complex terrain make collapse occurrence hard to forecast.
Key Innovation: The study applies a random-forest framework to map collapse susceptibility in Changping District and identify the main spatial controls on failure.
25. Long-term settlement of icy debris flow deposits: Process, underlying mechanisms and analysis model
Core Problem: Fresh debris-ice deposits can undergo large delayed settlement during thaw, but the deformation process and predictive treatment of this secondary hazard remain limited.
Key Innovation: Using the Sedongpu event as a benchmark, the study identifies a three-stage thaw-settlement mechanism in load-bearing ice-soil skeletons and develops an ice-content-dependent settlement model.
26. Dynamic response mechanism of mooring forces on a ship-type buoy under the coupled action of landslide-generated impulse waves and currents
Core Problem: Reservoir infrastructure design still lacks clear thresholds for how landslide-generated waves and currents load moored buoys and trigger tension failure or capsizing.
Key Innovation: The paper uses 3D physical model tests and wavelet analysis to quantify mooring-force dynamics and identify actionable failure thresholds for different buoy configurations.
27. Enhancing slope stability with biopolymer-cement composites
Core Problem: Slope stabilization still depends heavily on cement-based treatments with significant environmental costs.
Key Innovation: The study tests biopolymer-cement blends, identifies effective mixture ranges, and shows they can increase shear strength and modeled factor of safety while reducing cement reliance.
28. First Direct Observations of Internal Flow Structures in a Powder Snow Avalanche: Turbulence, Instability and Particle Distribution
Core Problem: Avalanche models still lack direct particle-scale field observations of airborne internal structure, turbulence, and instability.
Key Innovation: The paper reports the first high-speed optical field observations of particle motion inside a natural powder snow avalanche and links measured fluctuations to Kelvin-Helmholtz-type instabilities.
29. A review of field monitoring methods for bedrock frost weathering in cold alpine regions
Core Problem: Cold-region hazard research lacks a consolidated view of how bedrock frost weathering is monitored in the field and where the main observational gaps remain.
Key Innovation: The review synthesizes the evolution of field monitoring systems, monitored parameters, and unresolved measurement challenges relevant to frost-weathering-driven rock degradation and collapse hazards.
30. DRAD: A new model for Dynamic Real-time Avalanche Detection from videos with residual depth-separable convolution and feature pyramid networks
Core Problem: Real-time avalanche detection from video remains difficult when snow events must be separated from visually similar non-avalanche scenes.
Key Innovation: The paper develops a video-detection architecture that combines depth-separable convolutions, residual blocks, and feature pyramids for real-time avalanche recognition.
31. Optimised machine learning predictions of liquefaction-induced lateral spreading in alluvial deposits
Core Problem: Earthquake-induced lateral spreading in alluvial deposits remains difficult to predict accurately enough for resilient infrastructure design.
Key Innovation: The study systematically tunes and compares five supervised learning models, showing optimized CatBoost offers the strongest and most interpretable predictions.
32. Influence of topographic factors on the distribution of herbaceous vegetation and mechanical root properties of the Xijitan giant landslide in the upper reaches of the Yellow River, northwestern China
Core Problem: How topographic controls shape herbaceous vegetation patterns and root mechanics on giant landslides remains poorly quantified, despite clear implications for shallow stability.
Key Innovation: The paper links field vegetation surveys and single-root tensile tests to show how slope aspect organizes both plant composition and root strength on the Xijitan giant landslide.
33. A hybrid deep learning approach for highway landslide susceptibility assessment based on InSAR data
Core Problem: Highway landslide susceptibility studies still rely too heavily on static conditioning layers and underuse deformation-sensitive observations.
Key Innovation: The paper proposes a hybrid deep-learning susceptibility workflow that incorporates InSAR data to improve highway landslide assessment.
34. Sampling effects on machine-learning performance and tectonic controls on landslide susceptibility: insights from the Adra River basin (SE Spain)
Core Problem: Machine-learning landslide susceptibility models remain sensitive to sampling design, which can distort reported performance and obscure the tectonic controls of failure patterns.
Key Innovation: The paper develops an improved Adra River basin framework that tests sampling effects while resolving the tectonic signatures embedded in landslide susceptibility predictions.
35. Landslide recognition with sample augmentation based on a joint DCGAN and Pix2Pix
Core Problem: Automated landslide recognition is still constrained by limited labeled samples, which weakens detector generalization in mountainous terrain.
Key Innovation: The study uses a joint DCGAN-Pix2Pix augmentation strategy to expand training data and improve automated landslide recognition.
36. Fractal and wavelet diagnostics of flood resilience in transport corridors: Evidence from the 20-21 September 2025 Eastern Black Sea Floods, Türkiye
Core Problem: Mountain transport corridors need practical tools to identify where extreme floods most strongly degrade system resilience.
Key Innovation: The paper combines sliding-window fractal metrics with wavelet decomposition to diagnose fragile corridor segments after the September 2025 Eastern Black Sea floods.
37. Large-scale electromagnetic perturbations triggered during landslide events
Core Problem: Whether landslides generate robust electromagnetic precursor signals has remained uncertain because geomagnetic noise sources are difficult to isolate.
Key Innovation: The study filters major non-landslide noise sources, analyzes cross-station geomagnetic correlations, and argues that large-scale perturbations can emerge minutes before failure.
38. Village geohazard resilience assessment based on contribution weight superposition and Shannon entropy method- a case study in Dechang County, China
Core Problem: Village-scale geohazard resilience is rarely assessed with a framework that can directly support local mitigation and planning.
Key Innovation: The paper proposes a resilience assessment method that combines contribution-weight superposition with Shannon entropy for village-scale geohazard management.
39. Resilience assessment of geohazard emergency system based on the 'hazard-impact-drive' model
Core Problem: Emergency-system resilience for geohazards is still difficult to evaluate objectively while accounting for historical patterns, future trends, and internal drivers.
Key Innovation: The study introduces a hazard-impact-drive model to structure resilience assessment for geohazard emergency systems.
40. Energy relief effect of real-time drilling to prevent rockburst in high-stress rock
Core Problem: Preventive drilling for rockburst control lacks a clear energetic explanation of when and how it reduces failure intensity.
Key Innovation: The paper introduces a real-time drilling energy dissipation index and shows experimentally that larger drilling diameters reduce elastic strain energy, burst intensity, and rockburst proneness.
41. Investigating the liquefaction resistance and initial shear modulus of silty sands from equivalent intergranular void ratio concept
Core Problem: Liquefaction resistance and small-strain stiffness of silty sands remain difficult to interpret consistently because standard void-ratio concepts do not capture active fines participation.
Key Innovation: The paper fits equivalent intergranular void-ratio relations directly to test data and introduces fictitious active fines content to connect stiffness and liquefaction resistance.
42. Effects of non-plastic silt and soil aging on re-liquefaction resistance of sandy soils
Core Problem: The way non-plastic silt content and soil aging alter a sand's tendency to liquefy again after earlier liquefaction episodes remains poorly resolved.
Key Innovation: Repeated cyclic triaxial testing reveals a nonmonotonic fines-content effect on reliquefaction resistance and shows that aging-related advantages can persist partly after liquefaction history.
43. Spatiotemporal response of vegetation productivity to coseismic landslides: a case study of NPP dynamics following the 2014 Ludian earthquake, China
Core Problem: Post-event vegetation productivity response to coseismic landslides remains poorly characterized, limiting understanding of ecological disturbance and recovery after major slope failures.
Key Innovation: The paper uses MODIS net primary productivity time series to track the spatiotemporal vegetation response to landslides triggered by the 2014 Ludian earthquake.
44. Flow behavior and rheology of rock-ice granular mixtures with pendular liquid bridges
Core Problem: The influence of weak wetting and liquid bridges on the motion, segregation, and rheology of rock-ice avalanche mixtures remains poorly understood.
Key Innovation: Using DEM simulations with pendular liquid bridges, the paper shows how weak cohesion changes mixture integrity and segregation and finds that mu(I) rheology outperforms standard Voellmy-type formulations in the tested cases.
45. Response of the Sentinel-1 radar backscattering to an extreme wildfire event: surface soil moisture and vegetation cover implications
Core Problem: Post-fire hazard analysis still needs better remote-sensing evidence for how wildfire changes vegetation and near-surface moisture conditions that condition later slope failures.
Key Innovation: The paper links Sentinel-1 backscatter changes with Sentinel-2 vegetation indices, rainfall, and field soil moisture to characterize post-fire surface-condition shifts and recovery.
46. Seismic failure mechanisms of soil-structure-tunnel systems in liquefiable urban deposits: vector and scalar fragility assessment
Core Problem: Liquefiable urban deposits are still difficult to represent in coupled soil-structure-tunnel fragility analysis, especially when scalar intensity measures hide uncertainty.
Key Innovation: The paper develops validated finite-element simulations and vector fragility functions showing that multi-intensity-measure vulnerability estimates better capture liquefaction-driven damage.
47. Unified modeling on the coupled seismic dynamics of seawater − seabed − structure system via implicit VFEM
Core Problem: Seismic hazard analyses for liquefiable seabed foundations still underrepresent the coupled dynamics of seawater, seabed, and overlying structures, leading to underestimated failure potential.
Key Innovation: The paper develops a unified implicit VFEM algorithm that couples Helmholtz and Biot formulations to resolve seawater-seabed-structure interaction and its impact on liquefaction-driven marine damage.
48. GIS-based optimization framework for shelter site selection and population allocation under multi-hazard scenarios
Core Problem: Mountain communities exposed to both floods and landslides still lack integrated shelter-siting and population-allocation tools for emergency relocation.
Key Innovation: The study proposes a GIS-based optimization framework that jointly selects shelter sites and allocates population under multi-hazard scenarios in mountainous terrain.
49. Assessment of ground deformation in Mandalay, Myanmar, using InSAR with Sentinel-1 data after the March 2025 earthquake
Core Problem: Rapid post-earthquake deformation measurement in Myanmar has remained constrained by sparse ground observations and the need for fast synoptic assessment.
Key Innovation: The paper uses ascending and descending Sentinel-1 DInSAR observations to resolve vertical and east-west deformation after the March 2025 earthquake sequence and confirm the main seismogenic structure.