TerraMosaic Daily Digest: September 12, 2026
Daily Summary
Three studies address failure under sparse observation. A hierarchical Bayesian susceptibility model transfers information across regions while retaining local effects, outperforming conventional alternatives where landslide inventories are thin. TerrainGraphNet embeds slope continuity and elevation discontinuities directly into graph propagation for landslide delineation. For buried pipelines, mode-specific fragility surfaces resolve how landslide width and displacement redistribute local buckling, beam buckling, ovalization and tensile rupture rather than collapsing failure into a single threshold. Himalayan and landfill case studies then translate stability analysis into pile-anchor and leachate-well design constraints.
Process models connect subsurface structure to observable hazard. Thermo-mechanical simulations of central-vent calderas identify two ring-fault families, a transitional regime and a logarithmic relation between collapse underpressure and roof ratio. Gravity inversion and D-InSAR at Ethiopia's 2025 Fentale swarm show inherited lithospheric fabrics localizing magma intrusion and seismicity. GNSS analyses separate afterslip from viscoelastic relaxation after the Noto earthquake and resolve a J-shaped high-strain corridor across the southeastern Tibetan Plateau.
Hydroclimatic studies move from static maps toward time-dependent screening and routing. A quantum-classical generative framework ranks rare GLOF precursor anomalies while explicitly stopping short of deterministic forecasting. Coupled 1D-2D modelling identifies detention-area bottlenecks in the Daqing River network; Bayesian spatial correlation stabilizes rainfall-extreme estimation in data-limited Grenada; and temporal convolution optimized by CMA-ES extends flood routing across multiple lead times.
Monitoring and mitigation are increasingly tied to physical state. Seepage experiments and X-ray tomography show that particle migration reorganizes preferential channels and produces non-monotonic post-seepage strength. MICP slope covers reduce hydraulic conductivity and preserve suction during rainfall, while physics-informed microseismic localization adapts to changing tunnel geometry and sensor layouts. Socioeconomic evidence from Rwanda and flood-sensitive English housing markets further shows how physical exposure is converted into unequal livelihood and asset impacts.
Key Trends
The day's advances share a common aim: extracting defensible hazard state from sparse observations without discarding physical structure.
- Sparse-data transfer is becoming explicit: Hierarchical Bayesian pooling, terrain-conditioned graph reasoning and cross-scene remote-sensing transfer distinguish shared information from region-specific structure.
- Hazard models are resolving interacting failure modes: Caldera ring faults, landslide-pipeline damage and seepage-driven soil reorganization are modeled as competing or sequential mechanisms rather than single response variables.
- Geodetic observations are separating post-event processes: Horizontal and vertical GNSS constrain afterslip, viscoelastic relaxation and regional strain localization across earthquake and rift settings.
- Hydroclimatic prediction is moving toward probabilistic sequences: Rare-event generative screening, Bayesian rainfall extremes and temporal flood routing represent evolving forcing and uncertainty instead of relying only on static susceptibility.
- Monitoring is being designed around deployment constraints: Physics-informed microseismic localization, cloud-degraded satellite reconstruction and compact LiDAR processing target incomplete labels, changing sensors and operational latency.
Selected Papers
The September 12 collection is led by hierarchical Bayesian landslide susceptibility, thermo-mechanical caldera-collapse modelling and multi-sensor analysis of the Fentale rift swarm. Direct applications include terrain-constrained landslide segmentation, landslide-pipeline fragility, Himalayan and landfill slope stabilization, GLOF anomaly screening, rainfall-resistant bio-engineered slope covers, flood-network vulnerability and deep-tunnel rockburst monitoring. Companion studies advance earthquake deformation analysis, probabilistic rainfall extremes, flood routing, seepage-sensitive geomaterials and transferable Earth-observation methods.
1. Landslide Susceptibility in Data-Sparse Regions: A Hierarchical Bayesian Approach
Core Problem: Susceptibility models fail in regions with too few local landslides to estimate stable region-specific relationships.
Key Innovation: Pools information across regions through a hierarchical prior while retaining local parameters, improving prediction in data-sparse settings.
2. Thermo-Mechanical Controls on Central-Vent Caldera Collapse
Core Problem: Central-vent caldera failure lacks a unified mechanical explanation across roof geometries and thermal states.
Key Innovation: Couples magma dynamics, tectonic stress and nonlinear crustal rheology to identify two ring-fault modes and a scaling law for collapse underpressure.
3. The Role of Tectonic Inheritance in Active Rift Seismicity and Magmatism: Constraints From 3-D Gravity Inversion and D-InSAR Modeling of the 2025 Fentale Swarm, Ethiopia
Core Problem: The structures that localize magma intrusion and earthquake swarms during continental rifting remain poorly resolved.
Key Innovation: Co-locates a pressurized intrusion, low-density crustal volume and seismicity to show inherited lithospheric fabrics directing the 2025 Fentale swarm.
4. Mechanisms of Transient Postseismic Deformation Following the 2024 Mw 7.5 Noto Peninsula Earthquake Inferred From Horizontal and Vertical GNSS Data
Core Problem: Neither afterslip nor viscoelastic relaxation alone explains the horizontal and vertical GNSS response after the Noto earthquake.
Key Innovation: Jointly estimates the two mechanisms and resolves their contrasting near-field, far-field and time-dependent contributions.
5. Identification of flood routing characteristics and vulnerability in complex river networks
Core Problem: Complex detention networks require simultaneous simulation of inundation, routing time and infrastructure bottlenecks.
Key Innovation: Combines a validated 1D-2D model with flood marks, gauges and remote-sensing inundation to identify vulnerable reaches and control structures.
6. Seepage-driven structural reorganization governs hydro-mechanical evolution in channelized granular soils
Core Problem: Preferential seepage can reorganize granular soils in ways that decouple hydraulic evolution from later shear strength.
Key Innovation: Links seepage tests, X-ray tomography and direct shear measurements to gradient-dependent channel restructuring and non-monotonic mechanical response.
7. Multi-mechanism fragility surface assessment of buried steel pipelines subjected to landslide-induced permanent ground deformations
Core Problem: Single-mode fragility curves can underestimate pipeline failure where several deformation mechanisms compete spatially.
Key Innovation: Builds mode-specific and system-level fragility surfaces from nonlinear 3D simulations of landslide width and displacement.
8. TerrainGraphNet: Terrain-Constrained Graph Reasoning for Landslide Segmentation
Core Problem: Appearance-only segmentation fragments landslides and leaks across boundaries when spectral contrast is weak.
Key Innovation: Uses elevation-modulated features and slope-continuity graph edges to impose geomorphological coherence on landslide delineation.
9. Warnings must cross borders before disasters do
Core Problem: A cross-border mountain cascade can outrun fragmented detection, classification and authorization procedures.
Key Innovation: Frames warning performance around three clocks—preconditioning, physical propagation and institutional action—and requires end-to-end latency below hazard arrival time.
10. Harnessing machine learning and multi-source data fusion for regional geohazard risk assessment of a rapidly urbanizing mountainous region
Core Problem: Rapid urbanization requires hazard and vulnerability to be mapped together rather than treated as separate planning layers.
Key Innovation: Combines machine-learning susceptibility with game-theory vulnerability weights to isolate compact high-risk corridors in southeastern Sichuan.
11. Quantum generative intelligence for rare hydroclimatic anomaly screening of glacial lake outburst flood risk
Core Problem: Sparse GLOF records and subtle multivariate precursors make conventional supervised forecasting unstable.
Key Innovation: Uses quantum latent representations within conditional sequence generation and mismatch scoring to prioritize rare anomalies while preserving forecast uncertainty.
12. Influence of MICP on unsaturated soil properties and its application as bio-engineered slope covers on stability of slopes under varying rainfall conditions
Core Problem: Road-cut slopes need covers that limit infiltration without sacrificing unsaturated shear strength.
Key Innovation: Quantifies how microbial carbonate precipitation changes retention, conductivity and strength, then tests the resulting rainfall-stability response numerically.
13. Real-time microseismic source localization in deep tunnel using physics-informed neural networks
Core Problem: Microseismic localization must remain accurate as void geometry, velocity structure and sensor layouts change during excavation.
Key Innovation: Constrains travel-time learning with the eikonal equation and uses transfer learning for rapid adaptation to evolving tunnel conditions.
14. Vibration model test and seismic performance evaluation of polypropylene fiber reinforced concrete lining of mountainside tunnels in Loess Plateau
Core Problem: Mountain-adjacent loess tunnels experience coupled slope amplification and lining demand that conventional damping assumptions miss.
Key Innovation: Tests polypropylene-fiber-reinforced linings under shaking and quantifies strength gain and reduced bending response at critical sections.
15. Ground-Motion Spatial Correlation of Significant Duration of Horizontal and Vertical Components: A Case Study of 2021 MS6.4 Yangbi Earthquake in Yunnan, China
Core Problem: Regional lifeline analysis lacks spatial-correlation models for horizontal and vertical significant duration.
Key Innovation: Derives event-specific duration prediction and correlation functions from 311 Yangbi records and compares their decay with PGA and PGV.
16. Quantifying present-day strain rate of the southeastern Tibetan Plateau from high-resolution GNSS data: Implications for tectonic interactions and seismic hazards
Core Problem: Existing geodetic models lack the resolution needed to trace distributed deformation across major southeastern Tibetan fault systems.
Key Innovation: Adds 335 GNSS vectors and Euler-pole clustering to resolve a 600-km high-strain corridor linked to historical large earthquakes.
17. A Bayesian workflow for multi-station IDF curve development in data-limited island settings: application to Grenada
Core Problem: Short, incomplete island rainfall records cannot support spatially representative IDF curves through single-station analysis.
Key Innovation: Combines Bayesian spatial correlation, kriging and extreme-value models to generate coherent return-level fields across Grenada.
18. TCN-CMA-ES: a novel flood routing approach integrating temporal convolutional networks with covariance matrix adaptation evolution strategy
Core Problem: Flood routing models must preserve predictive skill across lead times without prohibitive calibration cost.
Key Innovation: Optimizes a temporal convolutional network with CMA-ES and evaluates discharge and stage forecasts across three river systems.
19. GIS-MCDA Based Flood Hazard, Vulnerability, and Risk Assessment in the Lower Nilwala Floodplain, Sri Lanka
Core Problem: Lower Nilwala planning requires consistent integration of physical hazard and social vulnerability at basin scale.
Key Innovation: Combines 22 indicators with AHP and weighted linear combination and validates separate hazard, vulnerability and risk surfaces.
20. Fluid-Driven Evolution of Deep In-Situ Stress Field in the Kuqa Depression: Insights From Seismic Monitoring and Stress Inversion
Core Problem: Fluid extraction and water breakthrough can redistribute stress through critically loaded faulted reservoirs.
Key Innovation: Integrates seismic catalogs, focal mechanisms, tomography and stress inversion to reconstruct westward stress migration during production.
21. The Impact of Landslide Disasters on the Socio-Economic Livelihood of Local Communities in Rwanda: A Case Study of Ngororero District (2014-2024)
Core Problem: Repeated slope failures in rural Rwanda affect livelihoods through pathways that physical hazard inventories alone do not measure.
Key Innovation: Combines household surveys and key-informant evidence to relate landslide occurrence to food insecurity, market disruption, income loss and recovery constraints.
22. Numerical Evaluation of Slope Stabilization Techniques for Residential Structures in Solan, Himachal Pradesh
Core Problem: A residential structure in the Lower Siwalik terrain occupies a slope with critically low static stability.
Key Innovation: Compares anchors, piles and combined reinforcement and shows that a pile-anchor system raises the modeled factor of safety to an acceptable level.
23. Influence of Varying Screen Length of Vertical Well Systems for Leachate Recirculation on the Slope Stability of Bioreactor Landfills
Core Problem: Pressurized leachate recirculation can concentrate pore pressure and destabilize bioreactor-landfill slopes.
Key Innovation: Quantifies how vertical-well screen length redistributes pore pressure and identifies a configuration that retains adequate stability across injection pressures.
24. Crustal Anisotropy and Deformation Mechanism of the Southern South China Block
Core Problem: Crustal deformation across the southern South China Block reflects multiple fault systems, subduction histories and mantle influence.
Key Innovation: Maps Pms splitting at 216 stations to distinguish region-specific fast-axis patterns and their tectonic controls.
25. Evaluating Changes in Convective Environments Over Subtropical South America Under Forced and Natural Climate Variability
Core Problem: Future changes in severe-convection environments are difficult to separate from large natural climate variability.
Key Innovation: Uses 50 climate simulations to quantify forced changes and uncertainty in instability, inhibition and vertical wind shear through 2100.
26. The Concept of the Mixing Index
Core Problem: Single-plume dilution measures do not capture mixing among separated waters with different transport histories.
Key Innovation: Derives a Jensen-Shannon mixing index and rate that handle multiple plumes, zero concentrations and anisotropic unsteady dispersion.
27. Representation strategies of nearshore short waves for predicting infragravity harbor oscillations: A systematic comparison
Core Problem: Operational harbor forecasts need compact short-wave representations that retain information about low-frequency resonance.
Key Innovation: Compares five spectral encodings and identifies compressed statistics that match full-grid prediction at substantially lower training cost.
28. TFSANet: Temporal Fusion and Structure-Aware Network for Remote Sensing Change Detection
Core Problem: Remote-sensing change models are vulnerable to pseudo-change and severe foreground-background imbalance.
Key Innovation: Fuses multidimensional temporal features and jointly optimizes pixel accuracy with region-level structural coherence.
29. Self-Supervised 3D Point Cloud Detection with Depthwise Separable Masked Autoencoders
Core Problem: Three-dimensional detection needs local geometric detail but annotated LiDAR data are expensive.
Key Innovation: Combines depthwise-separable masked attention, multiscale reconstruction and density-aware loss for label-efficient point-cloud learning.
30. Reconstructing cloud-free Sentinel-2 time series under complex degradations with state-driven spatio-temporal modeling
Core Problem: Cloud, blur and sensor noise can contaminate long satellite sequences when degraded observations are aggregated indiscriminately.
Key Innovation: Routes information from clear to degraded states and constrains attention by temporal succession and seasonality, with optional SAR support.
31. When multimodal fusion helps or hurts: Exploring modality synergy and cross-scene transferability for mangrove species mapping
Core Problem: More sensing modalities do not necessarily improve generalization when scene distributions shift.
Key Innovation: Evaluates all non-empty combinations of five modalities and uses contrastive domain adaptation to quantify synergy, redundancy and negative transfer.
32. Assimilating GRACE observations into a hydrological model in the Yangtze River Basin: The role of the Three Gorges Reservoir
Core Problem: Assimilation can misallocate reservoir mass changes to groundwater and soil stores when human regulation is absent from the model.
Key Innovation: Subtracts the Three Gorges Reservoir signal before ensemble assimilation and tests component consistency against independent hydrological observations.
33. Micro pore structure evolution of a loess-paleosol sequence and its control on macro hydraulic properties
Core Problem: Macroscopic permeability and collapsibility in loess-paleosol sequences vary sharply across buried microstructural interfaces.
Key Innovation: Combines oedometer, permeability, microscopy and porosimetry data to quantify how pore size and geometry control hydraulic and collapse response.
34. A path-driven structural evolution framework for SWRC hysteresis inspired by pore expansion and contraction processes
Core Problem: Conventional soil-water retention models poorly represent hysteresis and intermediate scanning paths.
Key Innovation: Adds a pore-evolution mechanism to ten established formulations and reproduces main and scanning curves across 20 soils.
35. Two-way coupled transient hydro-mechanical response around deep pressure tunnels in viscoelastic-plastic soils
Core Problem: Transient seepage and rheological deformation interact around deep pressure tunnels during startup and head changes.
Key Innovation: Extends consolidation theory with viscoelastic-plastic strain and two-way fluid-solid coupling for deep and twin shallow tunnels.
36. Surrogate-based sequential Bayesian updating of soil spatial variability for deep excavation-induced structural response prediction
Core Problem: Soil spatial variability creates evolving uncertainty in structural response as deep excavation progresses.
Key Innovation: Combines KL field reduction, a neural surrogate and sequential field-data updating to refine soil properties and serviceability predictions.
37. Flood risk information, housing markets, and spatial inequality
Core Problem: Nearby floods can alter housing transactions even without direct property damage, but the burden may differ by neighborhood deprivation.
Key Innovation: Separates price and time-on-market responses and finds larger implied risk-information discounts in more-deprived English neighborhoods.
38. Mathematical and Statistical Determination of Seismicity Index, Recurrence Interval and Seismic Risk of Shallow Earthquake Occurrence in Anatolian Fault in Turkey
Core Problem: Preparedness planning requires baseline estimates of shallow-earthquake frequency, recurrence and exposure-period probability along the Anatolian Fault.
Key Innovation: Fits Gutenberg-Richter parameters to 1,008 catalog events and derives magnitude-dependent occurrence rates and recurrence intervals.
39. Nonlinear seismic inversion in triaxial stress-induced anisotropic media modelled by microcrack closure mechanism
Core Problem: True triaxial stress closes microcracks anisotropically, complicating the recovery of stress and elastic properties from seismic amplitudes.
Key Innovation: Derives stress-dependent reflection physics and couples it with staged convolutional inversion for isotropic properties and anisotropy indicators.
40. Improving visibility of discontinuities on seismic data with colour processing technique: resolution concept across scales
Core Problem: Subtle faults and fractures remain difficult to trace consistently in conventional three-dimensional seismic volumes.
Key Innovation: Exports a structurally enhanced seismic volume that increases interpreted discontinuity density while explicitly documenting amplitude and sedimentary-feature limitations.
41. Adaptive Window-Width Trapezoidal Centroid Algorithm for High-Precision Pulsed LiDAR Ranging and Intensity Imaging
Core Problem: Fixed integration windows bias range estimates when LiDAR echo shape and reflectivity vary.
Key Innovation: Uses an adaptive trapezoidal centroid window to retain centimetre-scale ranging accuracy with hardware-compatible complexity.
42. DMAM: Dynamic Multiscale Adaptive Mechanism-Driven Remote-Sensing Target Detection Network
Core Problem: Dense small targets and large scale variation remain difficult under fixed sampling and uniform feature aggregation.
Key Innovation: Combines adaptive convolution, content-gated multiscale context and shape-aware regression across three aerial benchmarks.
43. How advanced information technologies help and hurt adaptation in crisis response supply chains
Core Problem: Information technologies can accelerate crisis response yet create new vulnerabilities under particular disruption patterns.
Key Innovation: Uses agent-based simulation and rule induction to distinguish conditional from broad technology-induced maladaptation.
44. Unified knowledge transfer boosts individual tree crown segmentation without scene-specific labels
Core Problem: Scene-specific labels limit cross-region and cross-resolution deployment of fine-scale segmentation.
Key Innovation: Aligns geographic domains, distills high-resolution structure into a low-resolution model and removes the need for target-scene annotation.
45. Anthropogenic nutrient loading masks topographic controls on geochemical patterns in a karst landscape: Implications for precision spatial zoning
Core Problem: Intensive nutrient inputs can obscure natural topographic controls on contaminant patterns in fragile karst landscapes.
Key Innovation: Combines a common geospatial grid with interpretable random forests to identify anthropogenic masking and nonlinear contamination breakpoints.
46. Root-Induced Soil Compaction (RISC): impact on hydraulic properties across spatial scales
Core Problem: The hydraulic consequences of root-induced soil compaction are poorly constrained beyond isolated roots.
Key Innovation: Links pore-scale observations, rainfall experiments and root-system simulations to show enhanced unsaturated flow and water retention.
47. Effects of different reclamation patterns on soil erodibility and soil nutrients in bauxite mining area of Guangxi, China
Core Problem: Bauxite-mine reclamation strategies require evidence on how they alter erodibility and nutrient recovery.
Key Innovation: Compares reclamation patterns to identify treatments that jointly improve soil resistance to erosion and nutrient status.
48. Numerical study on the optimal timing of bonded steel plate strengthening for shield tunnel
Core Problem: Bonded-steel reinforcement effectiveness depends on when strengthening is applied relative to damage development.
Key Innovation: Uses staged numerical analysis to identify the intervention timing that best limits shield-tunnel deformation and stress concentration.
49. A partitioned finite element formulation for large-strain sedimentation-consolidation of soft soils with application to multilayer deposition
Core Problem: Staged deposition requires sedimentation and large-strain consolidation to coexist across evolving material interfaces.
Key Innovation: Uses a partitioned finite-element formulation with stress-consistent interface mapping and deposition-driven layer activation.
50. A physics-informed neural network (PINN) framework for inverse mixed contact-traction tire-load reconstruction and finite-layer pavement response
Core Problem: Sparse observations do not uniquely reveal three-dimensional tire contact tractions and transient pavement response.
Key Innovation: Constrains inverse neural reconstruction with equilibrium, constitutive, interface and force-closure conditions under source-disjoint testing.
51. Performance-based selection of rubber-sand mixtures for buried pipeline backfill under static and cyclic loading
Core Problem: Sustainable pipeline backfill must balance static strength, cyclic deformation and energy dissipation.
Key Innovation: Uses static and cyclic triaxial tests to derive confinement-dependent selection criteria for rubber content and particle size.
52. Numerical Simulation of Seismic-Induced Damage Mitigation in Raft Foundations Using Scrap Tire Chips
Core Problem: Shallow foundations need low-cost damping layers that reduce settlement and acceleration under earthquake loading.
Key Innovation: Models a sand-tire-chip layer beneath a raft and reports substantial reductions in settlement and acceleration amplification.
53. Seismic Stability of Reinforced Walls with Construction and Demolition Waste and Quarry Waste Backfills
Core Problem: The seismic performance of reinforced walls constructed with recycled backfills is not established by conventional sand-based design assumptions.
Key Innovation: Compares quarry and construction-demolition waste under pseudostatic loading and quantifies their effects on sliding, overturning and reinforcement demand.