TerraMosaic Daily Digest: September 26, 2026
Daily Summary
Slope and coastal-flood studies accelerate assessment by learning from physical simulations. GeoRockSlope translates 494 finite-element simulations of Himalayan road cuts into static and seismic safety-factor estimates; its test accuracy measures reproduction of simulated stability, not independent prediction of slope failures. At two Brazilian beach profiles, Bayesian networks emulate XBeach storm-impact calculations, while reconstructed bathymetry and imagery-derived impact estimates address missing observations. The resulting predictions retain the site conditions and physical assumptions represented in their training simulations.
Earthquake studies distinguish source information from propagation effects and structural consequences. Coupled rupture-atmosphere simulations show that fault type and rupture speed affect thermospheric acoustic signals, with nonlinear propagation preventing direct interpretation of upper-atmosphere amplitudes as surface motion. Spectrum-conditioned generative modeling instead targets the engineering input, using an explicit consistency loss to improve synthetic accelerograms. Full-scale tests of timber-panel and friction-damper retrofits add experimental evidence on energy dissipation, while hospital isolation assessments distinguish collapse safety from acceleration-sensitive equipment protection.
Fractured-granite experiments resolve a non-monotonic coupling between infill thickness, permeability enhancement and elastic nonlinearity under dynamic stressing. Hydraulic changes therefore cannot be inferred from infill thickness without considering fracture roughness. At a different scale, a copula-based index represents dependencies among geological, physical and socioeconomic vulnerability components along the Kerala coast; comparison with historical tsunami impacts supports its local application.
Key Trends
Simulation surrogates, physics-constrained inference and consequence-specific validation connect the principal methodological contributions.
- Physics-based simulations train rapid surrogates: Finite-element slope calculations and XBeach storm simulations provide training targets for neural and Bayesian surrogates. Their predictive scope remains tied to the sampled geometries, forcing conditions and calibration evidence.
- Physical constraints enter source inference and signal generation: Coupled rupture-atmosphere modeling separates source effects from nonlinear acoustic propagation. For synthetic accelerograms, an explicit response-spectrum consistency loss reduces spectral residuals in event-independent tests.
- Seismic assessment follows component-specific consequences: Retrofit tests quantify hysteretic dissipation, hospital assessments track acceleration-sensitive equipment proxies, and a new intensity measure targets floor acceleration and velocity rather than collapse demand alone.
- Coupled variables replace isolated indicators: Fracture experiments relate hydraulic and elastic changes to infill thickness relative to roughness. Coastal vulnerability analysis uses copulas to retain nonlinear dependence among physical and socioeconomic components.
Selected Papers
Rock-slope safety-factor estimation, coupled earthquake-atmosphere dynamics and probabilistic coastal-impact prediction anchor this collection. Seismic experiments and life-cycle assessments connect ground-motion representation to retrofit performance and functional vulnerability. Complementary studies examine rock-mass geometry, rockburst, riverbank erosion and terrain reconstruction.
1. GeoRockSlope: an open access application for prediction of FEM-ANN guided road cut rock-slope instability
Core Problem: Road-cut stability in faulted Himalayan rock requires repeated static and seismic analyses across variable geometry and rock quality.
Key Innovation: GeoRockSlope uses 494 finite-element simulations to train surrogate models; optimized neural networks predict static and seismic safety factors for Baspa Valley road cuts. Reported test fits concern the simulation dataset, not independent regional failure forecasting.
2. Acoustic Wave Signatures of Earthquake Rupture Dynamics in the Upper Atmosphere
Core Problem: Upper-atmospheric acoustic signals cannot be interpreted as undistorted copies of earthquake surface motion.
Key Innovation: Three-dimensional coupled simulations distinguish rupture speed and fault type; modeled tsunami earthquakes produce acoustic amplitudes 20–50% below equal-moment thrust events. Nonlinear atmospheric propagation limits direct source inference.
3. A hybrid approach for rapid coastal flooding impact prediction in data-scarce environments: an early warning system perspective
Core Problem: Process-based overtopping calculations are expensive and require bathymetry and discharge observations that are often unavailable.
Key Innovation: Bayesian networks emulate XBeach simulations for two contrasting Brazilian beach profiles. Reconstructed profiles and non-professional imagery address data gaps, while probabilistic outputs describe site-specific impact uncertainty.
4. An integrated coastal vulnerability index using a copula-based approach: a case study of Kerala, India
Core Problem: Additive vulnerability scores can miss nonlinear dependencies between physical and socioeconomic conditions.
Key Innovation: A copula-based index combines fuzzy-normalized geological, physical and socioeconomic components along Kerala; comparison with 2004 tsunami impacts supports the dependency-aware formulation within this case study.
5. Generative AI for spectrum-compatible ground motion time history simulation
Core Problem: Synthetic accelerograms must match target response spectra without losing nonstationary waveform variability.
Key Innovation: An STFT-VAEGAN couples spectrum conditioning with an explicit response-spectrum consistency network. Event-independent tests and a Chi-Chi example evaluate spectral compatibility; ablation isolates the consistency loss contribution.
6. Life-cycle performance assessment of resilient base-isolated hospitals: collapse safety, structural performance, and functional continuity
Core Problem: Collapse avoidance alone does not quantify protection of acceleration-sensitive hospital equipment.
Key Innovation: Hazard-weighted life-cycle assessment compares single- and triple-friction-pendulum isolation using collapse, drift and floor acceleration. Floor acceleration is a functionality proxy, not a direct measurement of continued clinical operation.
7. Analysis of failure mechanisms in earthen buildings after the 2023 Al Haouz earthquake in Morocco and improvement solutions
Core Problem: Failure of rural earthen buildings reflects material, connection, foundation and site conditions that generic structural labels conceal.
Key Innovation: Field investigation following Morocco’s 2023 Al Haouz earthquake links observed failure modes with construction deficiencies and geotechnical conditions, informing recommendations for adobe, rammed-earth and stone construction.
8. Structural performance of RC frames retrofitted with CLT panels: experimental evidence from the e-CLT system
Core Problem: Pre-code reinforced-concrete frames require retrofit systems that add resistance without concentrating brittle damage.
Key Innovation: Full-scale cyclic tests evaluate cross-laminated timber panels with friction dampers. Retrofitted frames dissipate up to three times the unstrengthened frame’s energy near collapse-prevention drift, under the tested loading conditions.
9. A spectral-acceleration intensity measure for peak floor acceleration and velocity demands in super-tall buildings
Core Problem: Ground-motion scaling for collapse may inadequately represent floor acceleration and velocity demands on contents.
Key Innovation: An intensity measure calibrated across approximately 100,000 analyses is tested on a 489 m building archetype; stable average demands require seven records versus 23 under the compared PGA scaling, a case-specific efficiency result.
10. Poromechanics of Particle Remobilization and Interface Stiffness of Dynamically Stressed Fractured Rock
Core Problem: Dynamic stresses can change fracture permeability, but the role of infill thickness and interface stiffness is poorly constrained.
Key Innovation: Granite experiments combine pressure oscillations, permeability monitoring and ultrasound. Gouge thickness relative to roughness controls a non-monotonic hydraulic and elastic response; field fault-reactivation prediction is not demonstrated.
11. Spatial Assessment of Flood Susceptibility Using Remote Sensing and AHP-Based Multi-Criteria Analysis: A Case Study of the Mekerra Wadi Watershed (Algeria)
Core Problem: Flood-prone locations require a defensible combination of terrain, drainage, land-cover and rainfall indicators.
Key Innovation: The supplied RSS abstract describes AHP weighting and GIS overlay for Algeria’s Mekerra Wadi, with historical-event comparison and reported AUC 0.915. Bibliographic identity remains independently unverified.
12. A Prototype Geo-Digital Twin Framework for Scenario-Based Landslide Susceptibility Assessment Using Machine Learning
Core Problem: Susceptibility maps provide limited support for exploring environmental changes interactively.
Key Innovation: The supplied RSS abstract describes a Kentucky random-forest model embedded in Django and Mapbox, with NDVI perturbations and fuzzy interpretation. Scenario sensitivity is not causal vegetation-removal evidence or a validated real-time digital twin; bibliographic verification is pending.
13. Multi-stage seismic retrofit of buildings: a risk-based multi criteria decision-making procedure
Core Problem: Limited budgets require deciding which building stories to retrofit first.
Key Innovation: A risk-based greedy-search formulation is compared with a simplified TOPSIS prioritization on two buildings, using economic loss and structural demand measures to sequence retrofit stages.
14. Integrated Transformer approach for immersed-tunnel settlement prediction considering spatial heterogeneity and temporal degradation of sand filling compactness
Core Problem: Heterogeneous sand filling and its degradation complicate immersed-tunnel settlement forecasts.
Key Innovation: The corresponding author preprint combines elastic-wave characterization, settlement monitoring and compactness degradation in a Transformer. Journal identity is verified, but equivalence of precursor and final validation results is not established.
Method evidence: Author preprint / precursor; journal-version equivalence is not established.
15. Influences of tunnel radius on minimum overburden thickness for underwater shield tunnels crossing structural planes
Core Problem: Title-level focus: How do tunnel radius and structural planes affect the protective overburden needed above an underwater shield tunnel?
Key Innovation: Title-signalled approach: The title examines radius-dependent minimum overburden in discontinuous ground. Methods and results could not be assessed from a reliable abstract.
16. Influence of polyharmonic folds on rock-mass characteristics and slope stability in the Lidao-Wulu slate-schist contact zone, Taiwan: a scale-appropriate engineering geological model
Core Problem: Title-level focus: How should multiscale folding be represented when assessing slate–schist rock-mass stability?
Key Innovation: Title-signalled approach: The title links polyharmonic folds with rock-mass properties and slope stability in Taiwan’s Lidao–Wulu contact zone using a scale-appropriate geological model. Methods and results could not be assessed from a reliable abstract.
17. From Planning to Learning: Participatory Evaluation of a Volcanic Emergency Action Plan During an EU MODEX Exercise
Core Problem: Title-level focus: Emergency plans require evaluation of how participants actually learn and coordinate during exercises.
Key Innovation: Title-signalled approach: The title describes participatory evaluation of a volcanic action plan during an EU MODEX exercise. Methods and results could not be assessed from a reliable abstract.
18. Intelligent prediction under rockburst class imbalance: A coupled model fusing KMeans-SMOTE and CatBoost
Core Problem: Title-level focus: Rare severe rockburst classes can be underrepresented during model training.
Key Innovation: Title-signalled approach: The title couples KMeans-SMOTE resampling with CatBoost for rockburst prediction under class imbalance. Methods and results could not be assessed from a reliable abstract.
19. A hybrid smoothed particle hydrodynamics framework for modelling erosion in deformable riverbank
Core Problem: Title-level focus: Erosion models must accommodate changes in bank geometry and material deformation together.
Key Innovation: Title-signalled approach: The title proposes a hybrid smoothed-particle-hydrodynamics framework for erosion in a deformable riverbank. Methods and results could not be assessed from a reliable abstract.
20. Seismic performance of heavyweight quay wall backfill subjected to submerged conditions: Laboratory tests, numerical analysis, and 1g shaking table tests
Core Problem: Title-level focus: Submergence can alter the seismic response of heavyweight quay-wall backfill.
Key Innovation: Title-signalled approach: The title combines laboratory measurements, numerical analysis and 1g shaking-table tests to assess submerged backfill performance. Methods and results could not be assessed from a reliable abstract.
21. Stability Analysis of Embedded Cantilever Retaining Walls Using the Resistance Reduction Method
Core Problem: Prescribing a wall-collapse mechanism in advance can obscure competing soil and structural failures.
Key Innovation: Resistance reduction extends strength reduction to either soil or structural parameters in automated finite-element analyses, estimating required embedment without assuming the collapse mechanism.
22. CMD Net: Cross-Modal RGB-Depth Fusion for Dam Detection in High-Resolution Remote Sensing Imagery
Core Problem: Dams resemble other linear infrastructure and vary strongly in shape and size.
Key Innovation: The supplied RSS abstract describes CMD Net, gated texture–pseudo-depth fusion and a dam-image dataset, with regional detection tests. It maps dam objects rather than predicting dam failure; bibliographic verification remains pending.
23. Physics-Informed Liquid Neural Network Emulator for CRTM with Atmospheric-Layer Jacobian Capability
Core Problem: Fast satellite radiative-transfer emulators need physically meaningful vertical sensitivities as well as radiance agreement.
Key Innovation: The supplied RSS abstract combines layer-wise neural optical depth with a differentiable physical solver, reporting faster CRTM emulation and smoother Jacobians on a fixed 91-layer grid. Tests are clear-sky and absorption-only; bibliographic verification is pending.
24. A Hybrid Optimization Framework for Multi-Baseline InSAR Elevation Reconstruction Based on Sparse-Grid Quadrature Information Filtering
Core Problem: Phase noise and fringe boundaries impair interferometric elevation recovery.
Key Innovation: The supplied RSS abstract combines learned fringe boundaries, maximum-likelihood estimation and sparse-grid information filtering. This is elevation reconstruction, not demonstrated landslide deformation monitoring; bibliographic verification is pending.
25. Heatwave analysis in Iran: identification of regimes and scenario-based mapping of vulnerability
Core Problem: The location of heat vulnerability depends on both heatwave regimes and how exposure and social sensitivity are represented.
Key Innovation: Iranian heatwave regimes are combined with alternative vulnerability scenarios, revealing that vulnerable agricultural and water-dependent areas need not coincide with major urban centers. Its cross-hazard relevance is vulnerability methodology; it does not directly study geological hazards.
26. Electrical resistivity-based evaluation of water retention and gas permeability in MICP-treated granite residual soils
Core Problem: Nondestructive measurements are needed to track how calcite treatment changes unsaturated-soil transport.
Key Innovation: MICP-treated granite residual soils show increased water retention and lower gas permeability; electrical resistivity is evaluated as a proxy. Laboratory transport changes do not by themselves establish field slope stabilization.
27. Machine learning for modelling compositional soil properties in digital soil mapping
Core Problem: Soil fractions sum to a fixed total, making unconstrained Euclidean prediction and back-transformation potentially misleading.
Key Innovation: The same-title author preprint compares transformed and simplex-native random forests and reports the strongest performance for Dirichlet random forests. The journal DOI is verified, but the method summary is explicitly based on the preprint, replacing a mismatched raw abstract.
Method evidence: Author preprint / precursor; journal-version equivalence is not established.
28. Where does the eroded sediment from hillslopes go in small karst watersheds?
Core Problem: Hillslope erosion need not translate directly into exported catchment sediment in karst terrain.
Key Innovation: The author preprint highlights low sediment-delivery ratios in two karst catchments and attributes the disconnect to dual surface–subsurface drainage. These precursor findings are distinguished from the independently verified journal identity.
Method evidence: Author preprint / precursor; journal-version equivalence is not established.
29. GOVANS: A robust framework for automatic extraction of rock mass discontinuity orientations from 3D point clouds
Core Problem: Point-cloud edge effects bias surface normals, and ordinary clustering can poorly represent directional data.
Key Innovation: The corresponding author preprint describes robust statistics and global optimization for automatic discontinuity extraction. The journal identity is verified; quantitative journal-version validation remains unavailable.
Method evidence: Author preprint / precursor; journal-version equivalence is not established.
30. Microscale Hyperspectral Mapping of a Kilometer of Mantle Rocks: Insights Into Low-Temperature Serpentinization Systems From the Oman Drilling Program Cores
Core Problem: Discrete sampling can miss mineralogical heterogeneity along long rock cores.
Key Innovation: Submillimeter VNIR–SWIR mapping of one kilometre of Oman cores tracks hydration and iron-bearing mineral phases and identifies spectral ambiguities. The transferable contribution is continuous rock characterization, not demonstrated hazard prediction.
31. River Mouth Bar Formation and Evolution in Two Adjacent Estuaries Varying Waves and Estuarine Distance
Core Problem: Interactions between neighbouring estuaries complicate the relationship between wave forcing, jets and mouth-bar growth.
Key Innovation: Coupled Delft3D flow–wave experiments vary mouth spacing and wave forcing, separating deposition linked to velocity gradients from shear-driven resuspension and identifying distinct bar configurations.
32. Mixed grain boundaries mediate non-monotonic creep in hydrate-ice systems
Core Problem: Title-level focus: How do mixed grain boundaries affect time-dependent deformation in hydrate–ice systems?
Key Innovation: Title-signalled approach: The title identifies a non-monotonic relationship between mixed grain boundaries and creep; no field-instability outcome is inferred. Methods and results could not be assessed from a reliable abstract.
33. Geometry-aware dynamic convolutional network for instance-level segmentation of Martian landforms from high-resolution Digital Elevation Model-derived point clouds
Core Problem: Title-level focus: Instance-level terrain segmentation needs representations that preserve local three-dimensional geometry.
Key Innovation: Title-signalled approach: The title identifies dynamic convolution for Martian landforms in DEM-derived point clouds; application to terrestrial hazards remains prospective. Methods and results could not be assessed from a reliable abstract.
34. The overlooked role of plant hydraulics in shaping groundwater recharge projections
Core Problem: Title-level focus: Groundwater-recharge projections may be sensitive to how plant hydraulic processes are represented.
Key Innovation: Title-signalled approach: The title focuses on plant hydraulics as a control on projected groundwater recharge. Methods and results could not be assessed from a reliable abstract.
35. Evaluating the resilient behaviour of crushable subgrade soil
Core Problem: Title-level focus: Crushing changes subgrade response under repeated loading.
Key Innovation: Title-signalled approach: The title evaluates the resilient behavior of crushable subgrade soil; earthquake or slope applications are not established. Methods and results could not be assessed from a reliable abstract.
36. IDTS: An interpretable inertia-decay-aware temporal search framework for ultra-short-term offshore wind power forecasting
Core Problem: Using one look-back length for every variable can discard slowly varying information while retaining noise in faster variables.
Key Innovation: IDTS selects feature-specific windows with lightweight surrogate searches and transfers them to forecasting models, with SHAP-based lag interpretation and tests on two offshore wind farms. Applying this to rainfall–displacement monitoring is a prospective methodological use, not a demonstrated geohazard result.
37. Cascaded Dual-Radar Fusion Detection with Loose-Threshold Preliminary Screening and False Alarm Suppression
Core Problem: Variable sea clutter compromises detection thresholds and false-alarm control.
Key Innovation: The supplied RSS abstract describes loose-threshold fusion followed by pulse-continuity and frequency-stability discrimination on four sea-clutter datasets. Hazard-monitoring transfer is prospective; bibliographic verification is pending.
38. Proposal for using daily thermal amplitude to estimate reference evapotranspiration without relative humidity
Core Problem: Title-level focus: Reference evapotranspiration is difficult to estimate where humidity observations are missing.
Key Innovation: Title-signalled approach: The title proposes daily thermal amplitude as an alternative input when relative humidity is unavailable. Methods and results could not be assessed from a reliable abstract.