TerraMosaic Daily Digest: August 7, 2026
Daily Summary
Landslide studies resolve how pre-existing structure, changing water conditions and biological reinforcement govern failure across scales. At Wayanad, shear zones, joint orientations and lithological contrasts explain infiltration, sliding, valley confinement and exceptional debris mobility. In the upper Ganges, twelve years of imagery and InSAR reveal three instabilities coalescing into a 918,275 m² landslide dam with an approximately 178,000 m³ lake. At Danjiangkou Reservoir, swelling pressure reduces calculated safety factors by 18-32 percent during drawdown, while experiments on residual soil show that rainfall combined with rising groundwater produces the fastest disintegration. HARU-ASPP and Crack-YOLO extend this process focus into observation, improving landslide-boundary segmentation and converting UAV fissure masks into mine-slope fracture geometry.
Hydrometeorological and coastal studies separate forcing from the exposure and vulnerability that convert events into losses. Ural blocking and Rossby-wave breaking organized the atmospheric river behind Nepal's September 2024 disaster, which produced more than 450 mm of rain, 5.43 km² of inundation and 113 landslides. A Czech field survey identifies scour, debris impact and restricted openings across 122 bridges, including 20 collapses. In Xinjiang, hazard intensity explains only 42 percent of flood-risk variance because asset composition reverses basin rankings and regulation concentrates residual risk at breach corridors. Along Thailand's Andaman coast, scenario waves reach 26 m, while 82 percent of signposted evacuation routes are longer than the shortest alternatives, directly linking source physics to actionable preparedness.
Seismic and geodetic studies increasingly replace generic regional assumptions with observation-constrained models. CyberShake extends three-dimensional, simulation-based hazard estimates to 315 northern California sites and reduces bias against Loma Prieta records; a 72-branch logic tree built from 37,660 events refines hazard along the Sunda megathrust and Nicobar segment. Migrating Izu-Oshima earthquake swarms indicate that fractures opened during one fluid or magmatic episode can accelerate the next. Cargo-ship GNSS could supplement sparse tsunami gauges, while Bayesian geotechnical inversions and GPS-constrained fault modelling expose parameter, model-form and depth-dependent uncertainty rather than concealing it in a single best estimate.
Key Trends
Across the selected studies, the clearest methodological shifts concern process attribution, longer observational baselines, distributed sensing and explicit uncertainty.
- Hydrological pathways are being resolved inside slope-failure models: Rainfall, groundwater rise, reservoir drawdown, swelling pressure and root-modified water transfer are treated as distinct controls rather than collapsed into a single wetness proxy.
- Long records connect slope motion to cascading consequences: Decadal imagery and InSAR follow landslide growth, river blockage and lake formation, while event reconstructions link geological structure to runout, channel scour and temporary damming.
- Remote sensing is shifting from detection to measurable geometry: Attention-based segmentation preserves landslide boundaries, UAV instance segmentation retrieves fissure dimensions, and repeated topography quantifies post-disturbance sediment export.
- Hazard models are being tested against exposure and observations: Tsunami scenarios are paired with evacuation networks, flood intensity with asset vulnerability, and three-dimensional seismic simulations with recorded earthquake motions.
- Uncertainty is moving into the model comparison itself: Logic-tree seismic hazard, Bayesian finite-element calibration, layered CPTU inversion and fault-creep modelling distinguish source uncertainty, parameter uncertainty, model bias and depth-dependent mechanics.
Selected Papers
The strongest contributions combine process evidence with observations that can be revisited: structural mapping and hydromechanical monitoring for landslides, long-baseline deformation for river blockage, spatially explicit tsunami and seismic scenarios, field damage inventories for floods, and uncertainty-aware geotechnical inference.
1. Geological controls on large landslides: evidence from the 2024 Wayanad event, India
Core Problem: The structural and lithological controls that converted extreme rainfall into the unusually deep, mobile and destructive 2024 Wayanad landslide were unresolved.
Key Innovation: Field structure, petrography and XRF data link upstream-dipping infiltration pathways and downstream-dipping sliding planes to confinement, temporary damming, boulder transport and long runout.
2. Decadal evolution of debris flow-induced impoundment in the tributary of Pindar River, Upper Ganges, India
Core Problem: The growth and downstream hazard of a Himalayan landslide dam require observations that span more than a single failure episode.
Key Innovation: Twelve years of satellite imagery and SBAS-InSAR show three instabilities merging into a 918,275 m² mass, forming a roughly 178,000 m³ lake and accelerating by about 20 cm.
3. Hydromechanical behaviour and stability of reservoir landslides in the Danjiangkou Reservoir area under rainfall and water level fluctuation
Core Problem: Rainfall, reservoir drawdown and swelling pressure interact in expansive-soil landslides, but their separate contributions to progressive failure are poorly constrained.
Key Innovation: Monitoring, laboratory tests and coupled modelling show rainfall controlling shallow fissures, water-level change destabilizing the toe, and swelling pressure lowering safety factors by 18-32 percent.
4. Twenty years after the 2004 tsunami in Thailand with implications for future risk and preparedness
Core Problem: Thailand's Andaman coast remains exposed to megathrust tsunamis, but source scenarios and evacuation-route efficiency need joint assessment.
Key Innovation: Mw 7.8-9.4 simulations produce waves up to 26 m, while network analysis shows 82 percent of signposted routes exceed the shortest paths by an average 528 m.
5. Volcanic earthquake swarms at Izu-Oshima in 2023
Core Problem: The fluid and fracture processes that control migrating volcanic-earthquake swarms at Izu-Oshima remain uncertain.
Key Innovation: Waveform-correlation locations reveal southeastward diffusion in two 2023 swarms and a faster second sequence consistent with pathways opened during the first episode.
6. CyberShake Probabilistic Seismic Hazard Model for Northern California
Core Problem: Physics-based probabilistic seismic hazard estimates had not covered the full northern California San Andreas system at regional scale.
Key Innovation: CyberShake simulates hazard at 315 Bay Area sites and reproduces Loma Prieta observations with less bias than empirical ground-motion models and one-dimensional simulations.
7. Updated comprehensive probabilistic seismic hazard analysis of the Andaman-Nicobar and Sumatra regions
Core Problem: Seismic hazard estimates for Andaman-Nicobar and Sumatra require a unified catalogue, source representation and treatment of epistemic uncertainty.
Key Innovation: A 37,660-event catalogue, three source formulations, four weighted ground-motion models and a 72-branch logic tree resolve high hazard along the Sunda megathrust and Nicobar segment.
8. Evaluation of the ability of a cargo-ship GNSS network to detect tsunamis generated in the Solomon Islands region
Core Problem: Sparse offshore gauges delay tsunami detection in the southwest Pacific, particularly around the Solomon Islands.
Key Innovation: Simulations of the 2016 event and plausible scenarios show that moving cargo ships equipped with geodetic GNSS could detect tsunamis above Mw 7.0 despite vessel motion, heading and water-depth variation.
9. Role of atmospheric rivers and Rossby wave breaking in the September 2024 Nepal flooding and landslides
Core Problem: The large-scale atmospheric sequence behind Nepal's September 2024 flood-landslide disaster had not been resolved across weather, rainfall and inundation observations.
Key Innovation: ERA5, GPM and Sentinel-2 connect Ural blocking and Rossby-wave breaking to an atmospheric river, more than 450 mm of rain, 5.43 km² of inundation and 113 landslides.
10. Mechanism and control of toppling instability in anti-dip rock slopes reinforced by negative Poisson’s ratio anchor cables under mining-induced unloading
Core Problem: Deep mining unloads anti-dip rock slopes into large toppling deformation that conventional anchor cables may not accommodate.
Key Innovation: A constitutive DEM model and physical tests show negative-Poisson-ratio cables reducing maximum toppling displacement by about 70 percent while retaining large-deformation capacity.
11. Mechanism of lateral resistance in obliquely inclined bedding rockslides: a case study of the Shanyang rockslide
Core Problem: Lateral resistance in obliquely inclined bedding rockslides is difficult to infer from their deceptively modest apparent dip.
Key Innovation: UAV mapping, field and laboratory data, and 3DEC reconstruction of the Shanyang rockslide isolate the lateral-resistance loss that governs creep, bending and final instability.
12. Effect of rainfall and groundwater on disintegration pattern of granite residual soil in southeast China
Core Problem: Granite residual soil disintegrates under rainfall and groundwater change, yet tests under static water do not reproduce their coupled hydraulic pathways.
Key Innovation: Experiments distinguish blocky, layered and depression-style disintegration and show that rising groundwater combined with rainfall produces the shortest failure times.
13. Intensifying floods may not imply cascading flood risks in dryland rivers
Core Problem: Hazard intensification alone cannot explain why flood risk rises in some dryland basins but not others.
Key Innovation: Records, hydrodynamics and vulnerability modelling show hazard intensity explaining 42 percent of risk variance, while asset composition reverses basin rankings and regulation concentrates residual risk at breach corridors.
14. Hybrid attention residual U-Net++ with Atrous Spatial Pyramid Pooling (HARU-ASPP) for landslide segmentation
Core Problem: Landslide segmentation must retain fine boundaries while also representing multiscale context in heterogeneous remote-sensing imagery.
Key Innovation: HARU-ASPP combines residual learning, atrous spatial pyramids, dual attention and nested U-Net++ connections, reaching 80.45 percent mean IoU on Landslide4Sense.
15. An Empirical Express Method for Clay Slope Stability Assessment Based on Slip Surface Geometry and Factor of Safety Prediction
Core Problem: Preliminary clay-slope screening must account for water and geometry without the time required for a full numerical analysis.
Key Innovation: Empirical slip-circle geometry and safety-factor relations agree within 5 percent of Bishop analyses for most dry and groundwater-affected test cases.
16. Study on the Intelligent Recognition Algorithm for Open-Pit Mine Slope Fissures: Crack-YOLO with Texture and Semantic Enhancement
Core Problem: Manual interpretation cannot efficiently quantify small, blurred fissures across textured open-pit rock slopes.
Key Innovation: Crack-YOLO combines semantic and texture enhancement with instance segmentation, then converts masks into fissure length, width and azimuth; 196 fissures are mapped in about 22 seconds.
17. Integrated flood and multi-hazard susceptibility mapping in Egypt's Red Sea Mountains using AHP-machine learning, environmental sensitivity indices, and scenario-based restoration frameworks
Core Problem: Flash-flood mapping in arid coastal mountains is limited by sparse records, changing land use and the need to represent interacting environmental stresses.
Key Innovation: Sentinel-1 inventories, AHP-Random Forest and SHAP reach 0.95 AUC and connect flood susceptibility with environmental sensitivity and future climate-development scenarios.
18. Sea level rise risks for mega cities in the Yangtze River Delta
Core Problem: Future inundation and economic exposure of Yangtze Delta megacities under rising relative sea level and extremes remain insufficiently constrained.
Key Innovation: Observed sea-level trends and scenario simulations estimate exposed area, economic risk and embankment-overtopping thresholds.
19. Bridge and infrastructure damages following the September 2024 floods in the Czech Republic: post-disaster field survey
Core Problem: Bridge-damage models lack standardized field evidence from major floods across multiple structures and failure modes.
Key Innovation: A survey of 122 Czech bridges records 41 closures and 20 collapses, identifies scour, debris and hydraulic constriction as dominant mechanisms, and tests UAV classifications against field observations.
20. Incorporation of risk management into territorial planning: an analysis of the occupation of areas susceptible to floods and landslides in the State of São Paulo, Brazil
Core Problem: Risk maps do not reduce exposure if municipal land-use decisions continue to permit occupation of susceptible terrain.
Key Innovation: Across nine São Paulo municipalities, eight expanded into medium- or high-susceptibility flood and landslide zones after the 2012 law, revealing a gap between mapping and territorial control.
21. Soil reinforcement by Robinia pseudoacacia roots and its stabilization mechanism for shallow root-soil composite slopes
Core Problem: Root effects on shallow-slope stability are difficult to represent because hydrology, root architecture and reinforcement act together across potential slip surfaces.
Key Innovation: Field and shear tests calibrate analytical and numerical root-reinforcement models; anchorage reduces displacement by 73.5 percent and raises modelled safety factors by 180 percent.
22. Data-informed assessment of SSI effects on peak floor acceleration: a case study on the Fukushima nuclear power plants
Core Problem: Simplified seismic codes do not capture how deep embedment and high-frequency excitation alter floor acceleration in nuclear facilities.
Key Innovation: Fukushima records constrain empirical soil-structure transfer functions and a corrected analytical formulation, showing SSI can scale peak floor acceleration by factors of 0.2-0.8.
23. Revisiting flood resilience: How social-ecological interactions shape flood risk management policy and practice in Indonesia's riverbank urban slums
Core Problem: Technical flood metrics and lived risk perception can diverge in riverbank settlements, weakening the fit between formal plans and local action.
Key Innovation: Hydrological models, surveys, interviews and cognitive maps contrast collective warning in flash-flood-prone Bima with normalized, individualized coping under gradual flooding in Pontianak.
24. Multi-process assessment of soil-water-erosion dynamics in coir geocell-reinforced steep slopes under extreme rainfall
Core Problem: Extreme rainfall drives runoff, infiltration, erosion and strength loss simultaneously on steep engineered slopes.
Key Innovation: Large-scale tests show coir geocells reducing soil loss by 66.7-81.8 percent, increasing infiltration and suppressing preferential flow, rills and gullies.
25. Mechanical properties of soil reinforced by MICP and Eleusine indica: insights from grouting technique
Core Problem: Microbially induced carbonate precipitation can strengthen soil but may inhibit vegetation, leaving joint bio-reinforcement design uncertain.
Key Innovation: The experiments identify grouting conditions that preserve Eleusine indica growth while increasing cohesion and friction through carbonate cementation and root reinforcement.
26. Geomorphic changes and sediment yield dynamics in Alpine hillslopes: the role of surface temperature and rainfall intensity
Core Problem: Post-avalanche Alpine hillslopes lack paired measurements of topographic change, exported sediment and weather forcing.
Key Innovation: Repeated UAV surfaces and sediment traps show erosion exceeding deposition by more than 45 percent and link yield to land-surface temperature and rainfall intensity.
27. Spatiotemporal evolution of in-situ stress in active fault zones considering fault creep based on modified discontinuous deformation analysis
Core Problem: Planar fault models omit vertical stress gradients and therefore misrepresent the long-term stress consequences of creep at depth.
Key Innovation: GPS-constrained modified DDA embeds triaxial stress in two dimensions and resolves the transition toward tension, shear accumulation and depth-dependent fracture connectivity along the Jiali fault.
28. Experimental investigation and conceptual modelling of water-limited transpiration as hydraulic boundary condition in geotechnical engineering problems
Core Problem: Empirical reduction functions do not adequately represent how soil hydraulic conductivity, root architecture and atmospheric demand control water-limited transpiration in geostructures.
Key Innovation: High-suction planted-soil experiments support a closed-form physical reduction function linking transpiration decline to conductivity, root-length density and potential transpiration.
29. Pore segmentation and mechanical parameter prediction of water-infiltrated loaded sandstone based on improved Swin-UNet
Core Problem: Mesoscopic pore damage during cyclic water infiltration and loading is difficult to translate into macroscopic sandstone degradation.
Key Innovation: Improved Swin-UNet segmentation of in-situ CT images is coupled to porosity-based damage functions that predict strength and modulus loss.
30. Revealing spatial patterns and depth-dependent controls of soil organic carbon density across China using interpretable machine learning
Core Problem: China-wide soil-organic-carbon patterns and their depth-varying environmental controls are not adequately resolved by uniform empirical relations.
Key Innovation: Interpretable machine learning separates shallow controls dominated by temperature and vegetation from deeper controls associated with elevation and soil type.
31. The impact of climate change on the river basins of Parma (Italy) and Kamianka (Ukraine): a comparative analysis
Core Problem: Climate-driven vegetation and water-stress responses in contrasting European river basins are rarely compared within a common framework.
Key Innovation: A consistent NDVI, NDWI, temperature and precipitation analysis reveals stronger drought stress in Kamianka and stronger flood exposure in Parma.
32. Interpretable machine learning reveals spatial drivers and regional gradients of soil organic matter across Anhui Province, China
Core Problem: The spatial controls on soil organic matter across Anhui's climatic and topographic gradients remain poorly characterized.
Key Innovation: Tree-based models with SHAP expose nonlinear effects of nitrogen, precipitation, pH and terrain while producing a spatially explicit soil-organic-matter map.
33. Spatiotemporal changes and driving mechanisms of soil erosion in the Yellow River Basin: a case study of the Ten Tributaries
Core Problem: Long-term erosion decline and interacting drivers in the Yellow River's Ten Tributaries require a common explanatory framework.
Key Innovation: RUSLE, optimal-parameter Geodetector and partial least-squares structural equations identify persistent hotspots and strong interactions among slope, erodibility and human activity.
34. Spatiotemporal sediment connectivity in subtropical watersheds: Interactive effects of topography, vegetation, and rainfall
Core Problem: The relative and interacting roles of topography, vegetation and rainfall in watershed sediment connectivity are difficult to disentangle.
Key Innovation: Dynamic structural, functional and composite connectivity indices are tested against sediment yield and reveal seasonal shifts in dominant controls.
35. Machine learning-based soil depth estimation from the diverse landscapes of Nagaland, North-Eastern Himalayas
Core Problem: Soil depth is difficult to map across steep Himalayan terrain because field observations are sparse and environmental gradients are strong.
Key Innovation: Digital soil mapping compares multiple machine-learning models, identifies the best-performing estimator and quantifies the influence of precipitation seasonality and vegetation.
36. Assessing the spatiotemporal patterns and driving factors of water erosion in the Ganges-Brahmaputra-Meghna River Basin based on RUSLE-Geodetector
Core Problem: Water-erosion intensity and its regional controls across the Ganges-Brahmaputra-Meghna basin have not been consistently quantified through time.
Key Innovation: RUSLE and Geodetector reveal worsening erosion along the southern Tibetan foothills and a changing balance among topography, vegetation, precipitation and human activity.
37. Finite Element Bayesian Inference Using Dual Ensemble Kalman Filters for Uncertainty Reduction in Geotechnical Models
Core Problem: Bayesian calibration of nonlinear, high-dimensional geotechnical finite-element models remains computationally demanding and data-limited.
Key Innovation: Dual ensemble Kalman filters update material parameters and model states sequentially, reducing uncertainty from laboratory tests through a three-dimensional braced excavation.
38. Quantum Machine Learning for Infrastructure Risk, Resilience, and Disruption Management: A Review and Research Outlook
Core Problem: The practical maturity of quantum machine learning for infrastructure risk and resilience has not been critically established.
Key Innovation: A systematic review finds a narrow, mainly hybrid quantum-classical literature with limited empirical validation and identifies the benchmarks required before operational use.
39. Evaluation of an airborne polarimetric Geiger-mode lidar system
Core Problem: It was unclear whether airborne Geiger-mode lidar could add useful polarimetric discrimination beyond three-dimensional ranging and intensity.
Key Innovation: A calibrated evaluation finds consistent detection-probability differences across materials and transmitted and received polarization states.
40. Optimizing GNSS-derived terrestrial water storage anomaly retrieval: a spatial correlation accuracy crustal load inversion framework for enhanced hydrologic monitoring
Core Problem: Conventional GNSS terrestrial-water-storage inversion is too coarse to resolve local aquifer and sub-basin dynamics.
Key Innovation: A spatial-correlation-accuracy metric selects a 0.05-degree multi-resolution Green-function inversion that remains coherent with GRACE, GLDAS and precipitation.
41. Layered Bayesian model comparison and uncertainty quantification for CPTU dissipation test inversion
Core Problem: CPTU dissipation inversion confounds theoretical model bias with natural site variability when estimating consolidation parameters.
Key Innovation: Layered Bayesian comparison exposes systematic overestimation in simplified models and distinguishes the error, interval coverage and controlling uncertainty of competing formulations.
42. LIB4RT: Automated generation of reactive-transport formulations from PHREEQC databases for implementation in multiphysics platforms
Core Problem: Manual construction of reactive-transport formulations from geochemical databases is slow, error-prone and difficult to reproduce.
Key Innovation: LIB4RT parses PHREEQC databases into traceable reaction, speciation and transport formulations for multiphysics platforms and verifies them against benchmarks.
43. Numerical investigation of suffusion-induced infiltration patterns and microscopic interactions in gap-graded soils under cyclic seepage
Core Problem: Cyclic seepage drives fine-particle migration in gap-graded soil through pore-scale mechanisms that are difficult to observe directly.
Key Innovation: Coupled fluid-particle simulations distinguish retention, washout and clustering regimes and show how flow reversal reorganizes suffusion pathways.
44. Promoting microbial clogging through enhanced soil particle surface properties for improved seepage resistance in sandy soils
Core Problem: Microbial seepage barriers depend on organism attachment, which is inefficient on untreated sandy-soil particles.
Key Innovation: Coupling-grafted surface modification increases microbial adhesion and reduces hydraulic conductivity by nearly one order of magnitude.