TerraMosaic Daily Digest: June 13, 2026
Daily Summary
The June 13 issue is anchored by studies that treat landslides and debris flows as coupled geomorphic systems rather than isolated mapped polygons. The Mundakkai-Chooralmala landslide paper links slope failure, channel confinement, temporary damming, and downstream erosion into a single amplification framework. Along the Sichuan-Xizang corridor, stacked InSAR and spatial clustering produce a regional active-landslide inventory from ascending and descending Sentinel-1 observations. A Wumeng Mountain case study extends this monitoring line by adding winter thermal information to UAV photogrammetry, showing that land-surface temperature anomalies can distinguish crack connectivity, goaf influence, moisture contrasts, and deformation zones.
The strongest mechanics papers converge on unsaturated-to-flowing transitions. A unified elastoplastic-mu(I) model, a two-phase Riemann-based SPH formulation, and a variable-mass shallow-water SPH framework all address failure after the onset of large deformation, when pore pressure, entrainment, volumetric plastic strain, and runout stability dominate predictive skill. Complementary work on Qinghai debris-flow susceptibility links these process questions to future precipitation scenarios, while lidar-based sinkhole detection shows how repeated topographic surveys can convert hidden karst collapse into measurable rates. Tailings-storage InSAR and earthquake-corrected alpine-permafrost deformation analysis broaden this deformation-monitoring theme beyond natural slopes.
A broader methodological thread links geohazard monitoring to models that know when their observations are informative. Clay-CNN hybrids test whether geospatial foundation-model context improves landslide segmentation; self-supervised remote-sensing transfer and scaling-law studies quantify when general representations or additional labels help; and glacier-front segmentation, semantic change detection, selective SLAM fusion, DAS-based earthquake warning, and physics-constrained microtremor inversion extend the same question across cryosphere, infrastructure, seismic, and urban settings. The engineering papers translate this observation-first logic into liquefiable bridge foundations, tunnel-rock response, pile groups, ductile supports, soft soil stabilization, and post-disaster infrastructure restoration.
Key Trends
Five movements define the issue: process-coupled landslide mapping, explicit runout mechanics, tested remote-sensing AI, precursor-oriented monitoring, and infrastructure resilience modeling.
- Landslide science is moving from classification to process coupling: The Western Ghats, Sichuan-Xizang corridor, Wumeng Mountain, Qinghai debris-flow, and sinkhole papers all add deformation, thermal, topographic, geometric, or hydroclimatic controls to simple inventory mapping.
- Runout mechanics now targets the solid-fluid transition explicitly: Elastoplastic-mu(I), two-phase SPH, and variable-mass shallow-water SPH models focus on volumetric strain, pore-pressure evolution, entrainment, and numerical stability after failure begins.
- Remote-sensing AI is being stress-tested rather than assumed useful: Clay-CNN landslide detection, self-supervised transfer studies, scaling-law analysis, FreqChange, calving-front segmentation, and selective SLAM fusion ask when representation learning improves operational mapping.
- Monitoring is becoming precursor-oriented: InSAR deformation, UAV thermal anomalies, DAS amplitudes, microtremor-derived sediment thickness, lidar elevation differencing, tailings-storage subsidence, permafrost motion, and GNSS or SAR-derived fields provide measurable states before or during hazard evolution.
- Infrastructure risk models increasingly couple geotechnics with recovery: Liquefiable bridge foundations, segmental tunnel deformation, pile-group displacement, ductile tunnel supports, community resilience, HILP stress testing, and restoration optimization connect physical performance with system recovery.
Selected Papers
The selected papers emphasize coupled landslide-channel amplification, InSAR and UAV thermal monitoring, future debris-flow susceptibility, unsaturated-to-runout mechanics, tailings and permafrost deformation monitoring, sinkhole detection, earthquake early warning, remote-sensing foundation-model transfer, geophysical inversion, tunnel-ground interaction, and infrastructure resilience. This issue contains 40 selected papers from 1995 papers analyzed.
1. Beyond landslide classification: quantifying cascading hillslope-channel coupling and dam-amplified erosion during the Mundakkai-Chooralmala Landslide, Western Ghats, India
Core Problem: Extreme rainfall-triggered landslides in humid tropical mountains can amplify downstream damage through channels and temporary dams, but this coupling is rarely quantified.
Key Innovation: Integrates longitudinal-profile metrics, valley confinement, normalized steepness, chi analysis, a geomorphic amplification index, and field observations to formalize dam-amplified hillslope-channel coupling.
2. Regional-scale landslide detection along the Sichuan-Xizang transportation corridor using stacking InSAR and spatial clustering approaches
Core Problem: Sparse ground monitoring limits regional landslide inventories along the geologically active Sichuan-Xizang transportation corridor.
Key Innovation: Combines Sentinel-1 ascending and descending stacking InSAR with spatial density clustering to delineate 741 potential active landslides from deformation velocity fields.
3. Landslide Deformation Remote Monitoring in Alpine Mountains Using UAV Photogrammetry and Infrared Thermography: A Case Study in Wumeng Mountain Region, China
Core Problem: High-altitude winter landslides are hard to interpret from geometry alone because cracks, goaf zones, and moisture contrasts can produce different deformation signatures.
Key Innovation: Fuses UAV photogrammetry and infrared thermography to relate land-surface temperature anomalies to cracks, subsurface goaf connectivity, slumping zones, bedrock exposure, and moisture distribution.
4. Machine learning-based assessment of future debris flow susceptibility under CMIP6 precipitation scenarios in Qinghai Province, China
Core Problem: Climate-driven precipitation change may alter debris-flow hazard on the Qinghai-Tibet Plateau, but future susceptibility is rarely mapped with explicit climate scenarios.
Key Innovation: Uses environmental predictors, CMIP6 precipitation projections, and XGBoost to map present and future debris-flow susceptibility under multiple emission pathways.
5. A unified elastoplastic-mu(I) rheology model for unsaturated soils: formulations and two-point MPM implementation
Core Problem: Existing solid-fluid transition models often impose isochoric flow and static cohesion, limiting their ability to capture pore-pressure-controlled landslide motion.
Key Innovation: Couples Modified Cam-Clay and mu(I) rheology through stress decomposition, volumetric-yield evolution, structural degradation, and two-point material point implementation.
6. A two-phase Riemann-based SPH model for large deformation and rainfall-induced slope failure
Core Problem: Large-deformation rainfall failures require stable simulation of unsaturated soil-water coupling through both seepage and fluid-like runout.
Key Innovation: Adds a low-dissipation Riemann solver and unified transport-velocity treatment to soil-water SPH, validated against analytical solutions and rainfall-flume experiments.
7. A Variable-Mass Shallow Water SPH framework for debris flow simulation with erosion and entrainment
Core Problem: Conventional shallow-water SPH cannot naturally represent mass growth from bed erosion and entrainment during debris-flow propagation.
Key Innovation: Introduces variable-mass particles, a physically based erosion law, and two-way particle-bed coupling for depth-integrated debris-flow simulation.
8. Clay-CNN Hybrids: Leveraging Geo-Foundational Models as Auxiliary Context for Landslide Detection
Core Problem: Foundation models may not transfer directly to landslide mapping because failures occupy a small fraction of multispectral terrain chips.
Key Innovation: Evaluates Clay v1.5 on Landslide4Sense and shows that a U-Net augmented with Clay bottleneck context and LoRA outperforms Clay-only and U-Net baselines.
9. A novel approach to quantify cover-collapse sinkhole occurrences using multitemporal lidar
Core Problem: Karst sinkhole occurrence rates are hard to measure because collapses may form without visible precursors.
Key Innovation: Detects elevation changes between 2014 and 2023 lidar DEMs and filters non-sinkhole signals through multistep, aspect-based spatial analysis.
10. Localized ground deformation within the Aso caldera during the 2016 Kumamoto Earthquake: Combined effects of irregular basin geometry, soft lacustrine clay, and earthquake sequencing
Core Problem: Graben-type subsidence cracks in the Aso Caldera require explanation beyond peak shaking alone.
Key Innovation: Uses soil-water coupled elastoplastic simulations to show how basin geometry, sensitive lacustrine clay, foreshock weakening, and mainshock resonance jointly localized deformation.
11. DAS-based empirical amplitude prediction for earthquake early warning along Japan's Shinkansen railway
Core Problem: High-speed rail corridors need dense near-field earthquake monitoring, but conventional sensors are spatially sparse.
Key Innovation: Calibrates empirical amplitude prediction from distributed acoustic sensing along the Shinkansen corridor to support real-time railway earthquake warning.
12. Time-Series Monitoring and Analysis of Surface Deformation in Shiguilong Tailings Storage Using E-SBAS-InSAR
Core Problem: Tailings storage facility failures can cause severe casualties and losses, but dense multi-year deformation monitoring is difficult under complex surface conditions.
Key Innovation: Uses E-SBAS-InSAR and 88 Sentinel-1A images to retrieve 2022-2024 deformation, cross-checks SBAS-InSAR and PS-InSAR results, and decomposes seasonal responses to temperature and intense rainfall.
13. Seismic response of helical pile-supported bridge in liquefiable site
Core Problem: The seismic performance of helical-pile bridge foundations in liquefiable soils remains insufficiently quantified.
Key Innovation: Compares helical piles with reinforced-concrete piles and tests how helix size, number, and spacing control liquefied-soil deformation and bridge response.
14. Physics-constrained random forest for spatial prediction of shallow equivalent thickness from three-component microtremor data
Core Problem: Urban sediment thickness controls amplification but is difficult to map with sparse boreholes and unconstrained machine learning.
Key Innovation: Constrains random-forest prediction of shallow equivalent thickness with quarter-wavelength theory and three-component microtremor data.
15. Reversible Deep Operator Network for Grid-Independent, Multi-Scale Magnetotelluric Inversion and Uncertainty Quantification
Core Problem: Sparse and irregular magnetotelluric surveys often require retraining neural inversions when layouts, frequencies, or survey extents change.
Key Innovation: Combines RealNVP invertible networks with DeepONet to provide grid-independent forward and inverse MT modeling with uncertainty estimates.
16. How do Self-Supervised Remote Sensing Vision Models Transfer to Downstream Tasks?
Core Problem: The value of self-supervised remote-sensing encoders for downstream segmentation and regression remains inconsistent across tasks and decoder choices.
Key Innovation: Benchmarks six geospatial foundation models across multiple downstream datasets and shows that fine-tuning strategy and decoder capacity strongly control transfer gains.
17. The Scaling Law in remote sensing mapping for model performance forecasting and reference sample optimization
Core Problem: Large-scale mapping lacks quantitative guidance on how many reference samples are needed before model performance saturates.
Key Innovation: Formulates a log-logistic scaling-law framework to forecast performance limits, learning efficiency, and optimal sample allocation in remote-sensing mapping.
18. Multi-temporal calving front segmentation
Core Problem: SAR calving-front segmentation is degraded by seasonal ice mélange, snow cover, and frame-to-frame ambiguity.
Key Innovation: Processes multiple images in parallel and exchanges temporal features to stabilize calving-front delineation on the CaFFe benchmark.
19. Land Surface Deformation of Alpine Permafrost in the Earthquake-Impacted Source Area of the Yellow River During 2017-2024
Core Problem: Alpine permafrost deformation inferred from InSAR can be contaminated by co-seismic, aftershock, and post-seismic deformation in earthquake-affected regions.
Key Innovation: Uses SBAS-InSAR, earthquake-deformation simulation, and GNSS validation to remove the Mw 7.4 Maduo earthquake signal and isolate 2017-2024 permafrost land-surface deformation in the Source Area of the Yellow River.
20. FreqChange: Frequency-aware spatiotemporal learning with prior guidance for semantic change detection in remote sensing images
Core Problem: Bi-temporal change-detection models often miss changed regions or break semantic consistency by relying mainly on spatial-domain features.
Key Innovation: Adds frequency-aware temporal learning and semantic prior guidance to improve localization and class consistency in remote-sensing change detection.
21. Fuse only what matters: Degeneracy-aware multi-sensor fusion for LiDAR-Inertial-Visual SLAM
Core Problem: All-in sensor fusion can add noise and cost when visual, lidar, or inertial channels are not actually degenerate.
Key Innovation: Uses a selective Kalman-filter framework to detect sensor-degeneracy direction and fuse only the modalities that improve state observability.
22. MSRS-SLAM: A real-time multi-UAV dense mapping system for efficient low-altitude remote sensing
Core Problem: Single-UAV mapping is constrained by endurance and coverage during emergency response and resource surveys.
Key Innovation: Builds a multi-UAV dense reconstruction pipeline with visual odometry, deep multi-view depth, two-stage map fusion, and dense point-cloud registration.
23. Comparative study on seismic performance of aluminium, carbon steel and stainless steel moment-resisting frame buildings
Core Problem: The seismic behavior of aluminium moment-resisting frame buildings has not been systematically compared with conventional steel alternatives.
Key Innovation: Designs equivalent aluminium, carbon-steel, and stainless-steel frames and evaluates nonlinear pushover and dynamic response across building heights.
24. A Bayesian inference and machine learning based hybrid intelligence framework for modelling and assessing community resilience under uncertainty
Core Problem: Community resilience is shaped by interacting social, economic, and environmental variables that are difficult to assess under uncertainty.
Key Innovation: Integrates survey data, machine learning, and Bayesian belief networks to model uncertain resilience dependencies in coastal Khulna, Bangladesh.
25. Comparing U.S. household disaster preparedness across natural hazard risks: a Hierarchical Bayesian item response theory analysis of protective motivation and sociodemographic factors
Core Problem: Preparedness actions vary in difficulty and meaning across hazard types, making additive household preparedness measures misleading.
Key Innovation: Uses hierarchical Bayesian item-response modeling and a predictability-destructiveness typology to compare preparedness across five U.S. hazard-risk categories.
26. A scoping review and stress testing framework for High Impact Low Probability (HILP) events
Core Problem: Existing stress tests for high-impact low-probability events remain sector-specific and poorly matched to cascading infrastructure disruptions.
Key Innovation: Synthesizes stress-testing methods and proposes methodological elements for testing terrestrial critical infrastructures under uncertain extreme events.
27. A social-cost framework for prioritizing infrastructure restoration in post-disaster transportation networks
Core Problem: Transportation-network repair priorities often ignore welfare losses from delayed access to critical supplies.
Key Innovation: Combines congestion costs, deprivation costs, user-equilibrium assignment, and marginal vulnerability to rank restoration priorities after disruptions.
28. Planning ahead: Optimizing anticipatory action via mixed integer programming and open-source dashboards to mitigate climate risk
Core Problem: Anticipatory humanitarian aid must allocate scarce cash and supplies before hazards strike while respecting equity, budget, and capacity limits.
Key Innovation: Uses spatial mixed-integer programming, geospatial processing, and open-source dashboards to optimize anticipatory action for cyclone-prone Purba Medinipur, India.
29. Incorporation of cross-anisotropic small-strain stiffness into the Hardening Soil model
Core Problem: Isotropic small-strain assumptions can misrepresent soil-structure interaction where inherent cross-anisotropy controls stiffness degradation and recovery.
Key Innovation: Introduces cross-anisotropic small-strain stiffness into the HS-Brick refinement of the Hardening Soil model for finite-element simulation.
30. Assessing the durability of biopolymer-treated soil under weather cycles
Core Problem: Biopolymer soil treatment is promising for ground improvement, but its durability under wet-dry and freeze-thaw cycles is uncertain.
Key Innovation: Compares xanthan gum, sticky rice, lime, and untreated soils under weather cycles with microstructural and elemental analysis.
31. Analytical Solution for Multimode Deformation Behaviors of Segmental Tunnels Under 3D TBM-Lining-Foundation Interactions During Construction
Core Problem: Segmental tunnel linings under TBM construction are affected simultaneously by machine thrust, lining response, and foundation interaction.
Key Innovation: Derives analytical solutions for multimode deformation behaviors under three-dimensional TBM-lining-foundation interactions during construction.
32. Laboratory loading tests of ductile support systems under isotropic stress considering the effect of the installation position of yielding elements
Core Problem: Ductile supports are used in squeezing-rock mountain tunnels, but installation-position effects on support behavior remain experimentally underexplored.
Key Innovation: Uses scaled laboratory loading tests to evaluate yielding-element placement and mechanical response of ductile shotcrete support systems under isotropic stress.
33. Two simplified analysis methods for horizontal displacement of pile groups under tunneling considering pile head displacements synergy
Core Problem: Tunnel construction can displace adjacent pile groups, yet simplified methods often miss pile-head restraint synergy.
Key Innovation: Develops global-matrix and equivalent-stiffness methods that combine two-stage tunneling analysis, single-pile theory, and Mindlin solutions.
34. Effects of normal stress and shear rate on shear behavior of jointed rock mass bolted with high-strength high-toughness rock bolts
Core Problem: Conventional rock bolts can fail brittly under shear in faulted or highly jointed tunnel ground.
Key Innovation: Tests and simulates high-strength high-toughness bolts under direct shear to quantify strength-ductility gains and bolt-rock interaction.
35. Capillary force-driven migration of clay particles during wetting and drying of loess based on LBM-DEM
Core Problem: Fine-particle migration during wetting-drying cycles affects loess structure but is difficult to observe directly.
Key Innovation: Uses coupled lattice-Boltzmann and discrete-element modeling to simulate capillary-force-driven clay migration at gas-liquid interfaces.
36. An efficient high-order bound-preserving explicit-implicit-null time integration for the Richards' equation in unsaturated geotechnical media
Core Problem: Richards-equation simulations in geotechnical media are limited by sharp wetting fronts, nonlinear iterations, and severe time-step constraints.
Key Innovation: Combines high-order WENO spatial discretization with explicit-implicit-null time integration to preserve bounds while relaxing stability limits.
37. Tunnel post-blast half-hole recognition and compact structural consistency candidate re-ranking based on overbreak/underbreak flattened maps
Core Problem: Post-blast half-hole traces are weak, slender, and discontinuous indicators of contour control and rock-mass disturbance.
Key Innovation: Transforms point clouds into overbreak/underbreak flattened maps and reranks detector candidates using compact structural consistency.
38. UAV-based pavement distress detection and condition evaluation via divergent state-space modeling
Core Problem: Pavement cracks, potholes, and alligator cracking require scalable inspection without traffic disruption or manual survey cost.
Key Innovation: Develops a UAV imagery framework with efficient instance segmentation and divergent state-space modeling for multi-scale distress detection and condition evaluation.
39. Regional Climate Model Emulation with Diffusion Approaches: What is the Added Value of Generative Machine Learning?
Core Problem: Regional climate models are computationally expensive, and uncertainty-aware precipitation downscaling remains difficult for hazard assessment.
Key Innovation: Introduces and evaluates diffusion-based regional-climate emulators with ensemble uncertainty envelopes for high-resolution precipitation fields.
40. Extending the robustness analysis of uncertain minimum spanning trees via upper set tolerances
Core Problem: Infrastructure networks can fail when coordinated link deteriorations make an apparently optimal spanning tree suboptimal.
Key Innovation: Extends single-edge tolerance to upper-set and dual upper-set tolerances with polynomial algorithms for coordinated uncertainty in network robustness.