TerraMosaic Daily Digest: October 15, 2026
Daily Summary
Geometry and inherited ground conditions lead this issue's landslide studies. Nanfen's excavation simulations show why the retained rock domain matters when comparing displacement and damage between stages. A complementary Tibetan Plateau synthesis follows different failure types through to river blockage, bringing material sources and downstream pathways into the classification. Interbedded-rock creep tests and moisture-cycling slope experiments examine the conditions behind those broader patterns at controlled scales. Their practical value lies in identifying what changes within the rock or soil, as well as which parts of the terrain remain exposed to failure.
The flood papers make an equally consequential distinction between available data and dependable evaluation. TORRENT separates reported episodes, observation packages and a smaller curated benchmark for model comparison. Swiss probabilistic runoff experiments and global streamflow preprints then address uncertainty from different directions: better tail representation, the magnitude of ensemble spread and the decision rule used to issue a warning. Hanoi provides a concrete counterpoint: a model assembled under severe data constraints is calibrated against all available depth observations, so its reported agreement cannot serve as independent validation. These studies support more explicit links between observations, forecast distributions and decisions.
Changing boundary conditions also organize the coastal and seismic results. North Core Banks projections emphasize back-barrier submergence and repeated moderate events, while the Santee Sioux flood study finds localized connectivity changes rather than widespread intensification under its warming experiment. After geothermal shut-in, pressure and seismicity relax on different timescales. Dense Kīlauea observations resolve neighbouring magma-storage regions, and slow-slip models connect statistical signatures with fault interactions. Across these cases, the advances concern the structure, state and response of a system; translating them into event-specific warnings requires a further chain of evidence.
Key Trends
The papers connect evolving ground conditions with the observations and evaluation choices used to characterize hazard.
- Track the physical domain being compared: The rock remaining after excavation differs from the material present before cutting. This matters when interpreting modeled settlement and damage reductions.
- Keep the full landslide chain in view: The Tibetan classification distinguishes initiation pathways through to river blockage. Laboratory interbed and coal–rock tests add evidence about deformation partitioning and progressive failure within the source material.
- Distinguish data volume from benchmark independence: Large report inventories and numerous data packages do not necessarily provide equally many independent, well-observed events. Traceability and screening determine what a benchmark can test.
- Evaluate uncertainty magnitude as well as ranking: A useful ordering of forecast errors can coexist with under-dispersed ensembles. Learned warning decisions also depend on the costs encoded in rewards and on training stability.
- Use observations to constrain mechanisms at their measured scale: Post-injection pressure records, volcanic tomography and Antarctic grounding observations constrain different system properties. None directly supplies a universal timing rule for earthquakes, eruptions or ice instability.
Selected Papers
The selected papers move from mine slopes and high-mountain landslide chains to flash floods, coastal change and seismic processes. Field observations, controlled experiments and numerical reconstructions contribute different evidence. Forecast and mapping studies are assessed alongside the labels, calibration choices and validation data that support their results. Vector mapping, extreme-weather labels and fracture-characterization methods broaden the inputs available for exposure, forcing and subsurface assessment.
1. Three-dimensional progressive damage and geohazard assessment of an open-pit slope affected by subsurface voids and staged excavation: a case study of the Nanfen mine
Core Problem: Subsurface voids and excavation change both slope response and the rock domain being evaluated.
Key Innovation: A Nanfen model inherits damage across actual and planned excavation geometries. Lower settlement on the final retained surface partly reflects removal of material; stage-specific damage colours are not directly comparable magnitudes. The scenarios do not demonstrate healing of retained damaged rock or independently validated future safety.
2. Classification and Remote Sensing Characterization of Large Catastrophic Landslides in the Southeastern Tibetan Plateau, China
Core Problem: Large failures in high-mountain valleys need classification that includes the downstream chain.
Key Innovation: Field, optical, UAV, InSAR and environmental records distinguish rock–ice avalanche, debris-flow and rock-avalanche pathways to river blockage in the southeastern Tibetan Plateau. The synthesis supports process-aware prioritization, without validating a site-specific warning threshold or occurrence forecast.
3. TORRENT: An observation-constrained and benchmark-ready community shared dataset in the contiguous United States for flash-flood research and model intercomparison
Core Problem: Reported events, available observations and defensible evaluation cases need to be distinguished.
Key Innovation: TORRENT separates a 42,466-report inventory, 806 evaluation packages and 21 curated benchmark testbeds. The primary testbeds represent 14 episodes and are not a representative national sample. Traceable observations and documented screening support reproducible comparisons.
4. The need for uncertainty: why probabilistic LSTMs are key to improving flood predictions and enabling learned warning rules
Core Problem: Forecast distributions must be translated into decisions without hiding tail miscalibration or warning-cost assumptions.
Key Innovation: Swiss runoff tests compare probabilistic LSTMs and add the FRiCA warning-decision layer. Gains over fixed rules depend on reward design and random seed; the authors present an exploratory offline framework, not an operationally robust warning service.
5. KNOU-Flood: Advancing global streamflow forecasting with lead-aware probabilistic machine learning
Core Problem: Forecast inputs and ensemble construction affect how streamflow skill decays with lead time.
Key Innovation: KNOU-Flood tests a controlled LSTM extension across 2099 basins using matched inputs and splits. Lead-aware forcing and a probabilistic head improve evaluated skill, while benefits for rarer floods depend on training seed. The work extends an existing architecture and exposes protocol sensitivity rather than establishing universal extreme-flood skill.
6. Geophysical and microstructural signatures of crack genesis to failure in underground mining geosystems
Core Problem: Field seismicity and laboratory crack observations need a consistent account of progressive damage.
Key Innovation: Microseismic monitoring, acoustic emission, moment tensors and 4D micro-CT show converging descriptions of clustering, propagation and failure-mode changes. The fuller abstract updates the prior title-only record; correspondence across scales does not yet establish a transferable alarm threshold.
7. Failure Mode and Creep Characteristics of Soft–Hard Interlayered Rocks with Different Dip Angles and Layer-Thickness Ratios
Core Problem: Layer inclination and thickness ratios control how soft–hard rock combinations deform and fail.
Key Innovation: DIC/AE-monitored compression and shear-creep tests show changing tensile/shear localization and stress-dependent acceleration. The reported onset near 87% of peak strength belongs to the tested materials and loading paths, rather than a universal open-pit warning threshold.
8. Investigating Ice Aggradation and Thermal Erosion Induced by Seasonal Water Flow in Cold Regions With a Cold Room Flume
Core Problem: Seasonal surface flow can either build ice or erode frozen ground, depending on the heat and water supplied.
Key Innovation: Cold-room flume tests vary discharge and water/soil temperatures while tracking thaw and erosion. A transition near 1.3°C water temperature occurs under the tested conditions, with discharge modifying heat exchange. The threshold is experimental, not a universal air-temperature limit for permafrost instability.
9. High‐Resolution Body‐Wave Tomography of Kı̄lauea's Summit Magma Storage System From a Dense Passive and Active Source Experiment
Core Problem: Magma reservoirs require observations fine enough to distinguish neighbouring storage regions beneath a volcano.
Key Innovation: A dense passive–active experiment at Kīlauea combines more than 1800 nodal instruments and 396 source positions. Tomography resolves two shallow storage regions and is compared with geodetic sources. The result constrains reservoir structure, rather than the timing or probability of a future eruption.
10. Analysis of the Correlation Between Post‐Injection Wellhead Pressure Decay and Trailing Seismicity
Core Problem: Seismicity can persist after injection stops, complicating decisions about when a stimulation operation has relaxed.
Key Innovation: Multi-operation geothermal records relate seismicity-rate decay to wellhead-pressure relaxation following abrupt shut-in. Pressure relaxes more slowly in the analysed cases. The association supports a pressure-diffusion interpretation, without providing a universal restart threshold or guaranteeing the absence of later large events.
11. Flood Responses to Climate Warming on Tribal Lands Using High‐Resolution Hydrodynamic Modeling
Core Problem: National assessments can overlook local flood pathways and adaptation needs on Tribal lands.
Key Innovation: A 10 m hydrodynamic model for the Santee Sioux Nation uses paired historical and warming-adjusted runoff events. Results indicate localized changes in connectivity rather than widespread flood-extreme intensification. This site-specific outcome should not be generalized to all Tribal lands or all warming scenarios.
12. Exploring Future Ecogeomorphic State Transitions of an Undeveloped Barrier Island: Multidecadal Model Projections for North Core Banks, NC, USA
Core Problem: Barrier change may be controlled by back-barrier inundation as well as ocean-facing storm erosion.
Key Innovation: Probabilistic North Core Banks projections to 2050 track ecological and geomorphic states under sea-level and high-water uncertainty. Interior submergence dominates the modeled transitions, while frequent moderate events influence reorganization. Transfer depends on sediment supply and the characteristics of the barrier system.
13. A Simple Discrete Elastic Model Exhibits Both the Statistical Properties and Migration Patterns of Slow Slip
Core Problem: Elastic/frictional migration patterns and stochastic slow-slip statistics are difficult to reproduce within one simple model.
Key Innovation: A discrete elastic model combines locked/slipping elements with time-dependent strength recovery. It reproduces several migration and moment-rate signatures and connects them to a birth–death approximation. A remaining scale discrepancy in inferred slip dynamics limits interpretation as a complete subduction-zone model.
14. Effect of Initial Static Shear on Cyclic Softening Behavior of Very Soft Clay
Core Problem: Initial static shear changes how very soft clay responds to repeated loading.
Key Innovation: Undrained cyclic simple-shear tests examine pore pressure, stiffness and stress mobilization across shear and confinement conditions. An empirical pore-pressure model is compared with additional clay datasets. The results support condition-dependent design assessment, rather than a single monotonic rule for every measure of degradation.
15. Dynamic Response of a Suction-Caisson Foundation-Structure System through Numerical Simulations and Model Testing
Core Problem: Foundation capacity checks can overlook inertial interaction and the response of connected pipes.
Key Innovation: Centrifuge-supported nonlinear spring models are used in earthquake time-history analyses of a four-caisson manifold at Shah Deniz. The study follows structural motion and piping response, showing why equipment connections require assessment even when the main manifold remains robust.
16. Integrating DInSAR and Ice-Penetrating Radar for Assessing Grounding Line Consistency and Variations in Contrasting Antarctic Ice Shelves
Core Problem: Surface flexure boundaries need comparison with observations of the ice–bed interface.
Key Innovation: DInSAR and ice-penetrating radar are compared across contrasting Antarctic shelf sectors using 335 basal grounding points. Their spatial correspondence and differing retreat histories constrain grounding-zone interpretation. Agreement at surveyed lines does not eliminate temporal, positional or unsampled-sector uncertainty.
17. Effect of Roof Lithology on Deformation Coordination and Pre-peak Instability Evolution of Coal–Rock Combinations Under Uniaxial Compression
Core Problem: Roof strength alone does not describe how deformation and energy are partitioned before coal–rock failure.
Key Innovation: Layer-specific deformation, acoustic sources and energy measures distinguish coordination under different roof lithologies. Intermediate roof strength sustains greater storage and pre-peak deformation in these tests. Laboratory transition indicators require field testing before use as operational rockburst alarms.
18. Spatial assessment of wind–surge compound risk and vulnerability during Hurricane Ian: a GIS-based case study of Lee County, Florida
Core Problem: County-scale compound-hazard screening needs spatial checks against observed impacts.
Key Innovation: A GIS/AHP index for Lee County is checked against high-water marks and housing-damage records, with sensitivity tests for weights. The checks support the spatial zonation. A weighted relative-risk index is not a calibrated joint probability or a causal explanation of individual losses.
19. Review article: Mountain flash floods: from hazard formation to disaster risk reduction pathways
Core Problem: Fixed rainfall triggers overlook hydrological state, sediment connectivity and the conditions turning a flood into a disaster.
Key Innovation: A synthesis of 1967 studies combines bibliometric patterns with process and governance evidence. It connects antecedent state, hillslope–channel pathways and coordinated mitigation portfolios, while highlighting geographical imbalance and the limits of transferring fixed thresholds.
20. Modality-Semantic Reliability in VLM-Based Temporal Landslide Explanation: A Rule-Constrained and DEM-Aware Alignment Framework
Core Problem: A static DEM cannot establish terrain change between image dates, yet generated explanations may attribute such claims to it.
Key Innovation: A DEM-aware alignment framework reduces several task-specific misuse measures. On 200 blinded expert-reviewed pairs, misuse decreases from 95.5% to 87.5%, leaving substantial residual error. This fuller evidence updates the earlier title-only entry; schema consistency and narrow rule scores must not be equated with geologically reliable explanations.
21. Ranked Right, Sized Wrong: Ensemble Spread Fails Silently at Streamflow Extremes
Core Problem: An ensemble can rank errors well while reporting a severely inadequate uncertainty magnitude.
Key Innovation: Streamflow ensembles show widening error-to-spread discrepancies at high flows, especially beyond the training record. Near-oracle uncertainty rankings therefore do not establish calibrated flood probabilities. The reported ratios describe the evaluated ensemble and conditioning choices, not a universal correction factor.
22. Building and calibrating a 3Di subgrid flood model for Hanoi, without a surveyed terrain model, drainage data, or rain gauges
Core Problem: Urban flood reconstruction is difficult without surveyed terrain, drainage records or rain gauges.
Key Innovation: A Hanoi 3Di model repairs terrain and uses satellite rainfall and land-cover infiltration as drainage proxies. All eight depth observations enter calibration; two sites remain dry. Agreement at the other sites is calibration fit, not independent validation of predictive depth accuracy.
23. SAR Coherence-Guided Machine Learning Framework for Flood-Induced Damage Mapping using a Single Fully Polarimetric SAR Scene
Core Problem: Coherence-derived labels and modeled flood depths are imperfect proxies for physical damage.
Key Innovation: An ALOS-2 polarimetric framework maps the Hagibis event and compares spatial patterns with FwDET depths. This extends the earlier title-only entry. The depth-based check does not independently validate structural damage, and vegetation, urban scattering and cross-event transfer remain limitations.
24. Can AI Forecast the Extremes? Evaluating Probabilistic Weather Models on Tropical Cyclones
Core Problem: Average weather skill can hide shortcomings in cyclone track, intensity and ensemble uncertainty.
Key Innovation: GenCast, FourCastNet 3 and FGN are compared with IFS-ENS using probabilistic diagnostics. Better location scores coexist with intensity and spread limitations. The reported tracking advantage is specific to the evaluated storms and metrics, rather than evidence of uniformly superior cyclone warnings.
25. Digital Typhoon Dataset V3: Benchmarking for Multi-spectral Extension and Analyzing Systematic Bias in Intensity Regression
Core Problem: Long satellite records need consistent evaluation of spectral information and intensity-regression bias.
Key Innovation: Digital Typhoon V3 adds multispectral data and examines split choices, targets and hemispheres. More channels do not consistently improve regression, and imbalance weighting does not remove strong-storm intensity bias. The dataset supports hazard monitoring research while documenting persistent benchmark limitations.
26. MARS - A foundational map auto-regressor
Core Problem: Pixel segmentation followed by vectorization can compound errors when extracting roads and building polygons.
Key Innovation: MARS generates both road polylines and building polygons autoregressively and releases a 3.4-million-example training resource. Four extraction benchmarks and interactive correction demonstrate a substantial mapping approach. Disaster damage, disrupted access and exposure accuracy still require direct evaluation.
27. Seismic Imaging of the Crust From Long‐Offset Ocean‐Bottom Seismometer Data Using Time‐Domain Extended‐Source Full‐Waveform Inversion: Application to the Eastern Nankai Trough, Japan
Core Problem: Long-offset waveforms can drive crustal inversion toward cycle-skipped solutions.
Key Innovation: Extended-source full-waveform inversion is tested on eastern Nankai ocean-bottom seismic data, followed by conventional refinement. Improved data fit and resolved fault structures support the imaging workflow; source geometry alone does not determine earthquake occurrence or tsunami likelihood.
28. A Generalized Unit Hydrograph Theory for Dynamic River Networks
Core Problem: A fixed river network cannot describe all changes in catchment response as channels expand and contract.
Key Innovation: A generalized unit-hydrograph convolution adds a term for changing active stream length. Numerical tests and one monitored catchment event support effects on peaks and recessions. The field application is proof of concept, with broader calibration and transfer still required.
29. Transboundary fire regimes in the sino–mainland southeast asia borderlands
Core Problem: Administrative boundaries can divide connected fuel landscapes and obscure asymmetric cross-border exposure.
Key Innovation: Satellite fire events in bilateral buffers reveal differences in counts and exposure across China–Myanmar, China–Laos and China–Vietnam borders. Similar event-size and spread characteristics coexist with these contrasts; the associations do not isolate the causal effect of national fire policy.
30. Ground-Deconvolved GEDI Waveforms Reveal Ladder Fuel Structure in Open Dry Dipterocarp Forest: A Ground-to-Space Sensor Comparison
Core Problem: Different laser and optical sensors resolve different parts of an open forest's fuel profile.
Key Innovation: Ground, UAV and satellite comparisons at Sakaerat identify stronger evidence for ladder fuel than near-surface material. Ground-return deconvolution improves a GEDI signal, while sparse two-mode detections limit regional extrapolation. The measured fuel fractions do not directly estimate ignition or fire spread probability.
31. A Novel Wellbore Nitrogen Leakage Testing Model for Salt Cavern Compressed Air Energy Storage: A Case Study
Core Problem: Pressure stabilization or a stationary gas–liquid interface does not necessarily establish sealing integrity.
Key Innovation: A mass-based nitrogen-leakage model interprets field observations from Tai'an compressed-air storage and identifies an interwell pathway. Temperature-driven pressure cycles are separated from leakage interpretation. The site's engineering assessment does not establish a general absence of storage leakage risk.
32. Infrared Brightness Temperature Evolution Modulated by Pressure-Stimulated Rock Potential in Diorite Partly Loaded to Fracturing
Core Problem: Electrical and infrared changes during rock loading require separation from temperature and background radiation.
Key Innovation: Partly loaded diorite experiments relate pressure-stimulated potential and infrared evolution to proposed charge processes. The work concerns laboratory/outdoor fracture sensing; it does not validate earthquake prediction from satellite thermal anomalies.
33. Near-Surface Water Redistribution, Deformation, and Runoff Responses of Expansive Soil Slopes with a Biochar–Bagasse Fiber-Amended Surface Layer Under Rainfall–Evaporation Cycles
Core Problem: A surface amendment can change moisture redistribution as well as measured slope movement.
Key Innovation: Two slope models undergo five rainfall–evaporation cycles, with lower recorded displacement in the amended model. The comparison describes these models and preparation conditions. It does not establish a replicated field-scale reduction in landslide probability or long-term performance.
34. Phreatic line revision and prediction for dry-stack tailings storage facilities based on BP neural network
Core Problem: The verified study revises and predicts phreatic lines in dry-stack tailings storage using a neural network.
Key Innovation: Not assessed from the available bibliographic record.
35. A Catastrophe Modelling Approach for Rapid Post-Event Flood Loss Estimation
Core Problem: The verified study develops a catastrophe-modelling approach for rapid post-event flood loss estimation.
Key Innovation: Not assessed from the available bibliographic record.
36. A Novel Nexus of International Migration, Land Use Change, and Disaster Vulnerability: A Structural Equation Modeling (SEM) Analysis for Kerala, India
Core Problem: The verified title examines migration, land-use change and disaster vulnerability through a Kerala case study.
Key Innovation: Not assessed from the available bibliographic record.
37. Seismic resilience assessment of water distribution systems considering improved fragility relations and CIPP rehabilitation
Core Problem: The verified study evaluates water-distribution seismic resilience with revised fragility relations and CIPP rehabilitation.
Key Innovation: Not assessed from the available bibliographic record.
38. Stochastic longitudinal model of buried pipelines considering soil spatial variability and void geometric uncertainty
Core Problem: The verified study models buried-pipeline response with uncertain void geometry and spatially variable soil.
Key Innovation: Not assessed from the available bibliographic record.
39. Deterioration mechanisms and safety performance evolution of tunnel linings under coupled high-temperature groundwater and sulfate attack
Core Problem: The verified study examines tunnel-lining deterioration under high-temperature groundwater and sulfate exposure.
Key Innovation: Not assessed from the available bibliographic record.
40. A dual-criteria procedure for mainshock-aftershock sequence generation: Significant duration and conditional spectrum
Core Problem: Synthetic earthquake sequences should preserve duration as well as spectral characteristics.
Key Innovation: A duration- and conditional-spectrum procedure selects mainshock–aftershock pairs and tests a case-study structure. Aftershocks increase hysteretic energy even when deformation changes little. The findings remain conditional on the selected records, hazard levels and structural model.
41. Hydro-thermal processes and frost heaving characteristics of unsaturated expansive soil under unidirectional freezing conditions
Core Problem: The verified study examines hydrothermal processes and frost heave under unidirectional freezing.
Key Innovation: Not assessed from the available bibliographic record.
42. Temporal and Spatial Dynamics of Critical Shear Stress of Bedload Transport in Gravel‐Bed Rivers Under Variable Sediment Supply Conditions
Core Problem: A fixed critical shear stress can miss the effects of sediment-supply history on channel transport.
Key Innovation: Eight gravel-bed experiments show changing inferred transport thresholds during bed adjustment, including longitudinal differences associated with deposition and slope. The results inform sediment routing under non-equilibrium supply; they are not a direct validation of debris-flow or flood-hazard predictions.
43. DIAL: Dynamic and Informed Adjustment of the Native Stochastic Law for Reliable Weather Ensembles
Core Problem: Weather-ensemble spread can be miscalibrated at particular scales and states despite reasonable global averages.
Key Innovation: DIAL adjusts a frozen model's native stochasticity during rollout and is tested on later ERA5 years. Improved multiscale probabilistic scores support the control strategy, but hazard-specific reliability and realized flood-warning value are not evaluated.
44. TailWeather: from tail to extremes, a global climatological dataset for machine-learning weather forecasting
Core Problem: Climatological rarity and documented disaster impact are different targets for evaluating forecasts.
Key Innovation: TailWeather labels ERA5-derived tails consistently over land and compares them with reported impacts. Reporting gaps limit local impact validation. The product supports extreme-event benchmarking; a rare-weather label is not itself a damage observation or calibrated disaster probability.
45. Stereological invariance in 3D fracture characterization: distinguishing natural fracturing from superimposed anthropogenic disturbances
Core Problem: The verified title distinguishes natural fracture patterns from superimposed anthropogenic disturbance in three-dimensional characterization.
Key Innovation: Not assessed from the available bibliographic record.
46. A workflow for using Normalized Correlation Count to characterize 1D fracture spatial arrangements
Core Problem: Statistics computed across non-stationary fracture scanlines can misrepresent spatial structure.
Key Innovation: A workflow uses cumulative spacing and its derivative to identify stationary domains before interpreting fracture spacing and saturation. This supports geological characterization; stability or flow predictions still require physical parameters and site-specific models.
47. A knowledge-guided machine learning framework for reconstructing climate-driven subsurface water dynamics from time-lapse electrical resistivity tomography
Core Problem: Temperature effects must be separated from resistivity changes used to interpret subsurface wetting.
Key Innovation: A knowledge-guided model reconstructs monitored hillslope resistivity and tests warming scenarios. Results concern a site-specific water-content proxy within the observed record, rather than direct pore-pressure, slope-stability or early-warning prediction.
48. Cross-scale reconstruction of rock discontinuity rough surfaces: A waveform-based extrapolation framework
Core Problem: Limited outcrop exposure constrains rough-surface models across scales.
Key Innovation: An author-group record describes cosine-transform and random-phase reconstruction, including anisotropy and upscaling tests on Beishan granite discontinuities. This supports geometry generation; field-scale failure or flow performance is not demonstrated by the surface-reconstruction result alone.