Initiated by Dr. Xin Wei, University of Michigan
Ongoing development by the community

TerraMosaic Daily Digest: Mar 21, 2026

March 21, 2026
TerraMosaic Daily Digest

Daily Summary

This March 21, 2026 digest distills 40 selected papers from 951 analyzed records. The clearest pattern is a turn toward latent-state estimation in hazardous systems. Landslide-related studies now infer physically meaningful internal variables from sparse observations: microseismic classifiers separate fracture-related events under single-station conditions, mineralogical enrichment is converted into a proxy for residual shear strength in Himalayan thrust-zone soils, and sediment-starved Yangtze reaches are linked quantitatively to intensified bank erosion and collapse.

A second cluster centers on coupled underground failure. Rockburst incubation, mining-induced seismicity, tunnel-face stability, fracture seepage, acid stimulation, and grouted support defects are all treated as flow-stress interaction problems rather than isolated strength exceedances. Hydrological papers echo the same logic: physics-informed hillslope subsurface flow, process-aware urban drainage surrogates, karst and lagoon monitoring, and threshold-based nitrogen mobilization all show that the most useful models today are not merely predictive, but mechanistically legible.

Key Trends

Today's strongest studies convert hard-to-observe internal states into usable hazard indicators, tightening the link between monitoring, mechanics, and operational decision-making.

  • Slope-hazard work is shifting inward: microseismic signatures, mineralogical weakening, suspended-sediment deficits, and climate-conditioned reliability are being used to infer the internal evolution of unstable slopes.
  • Underground engineering is dominated by hydro-mechanical interaction: rockburst, tunnel-face stability, fracture permeability, acid stimulation, and support defects are all framed through coupled stress–flow processes.
  • Monitoring pipelines are becoming more interpretable: cave drip chemistry, lagoon time series, river optics, and point-cloud discontinuities are no longer just observations; they are being turned into process-specific indicators.
  • Environmental extremes are being analyzed structurally: papers on heat, floods, precipitation, and nitrogen export emphasize routing, temporal structure, and thresholds rather than only aggregate intensity.
  • Mechanics-aware machine learning is strengthening: gray-box denoising, active-learning slope reliability, and process-aware graph surrogates illustrate a clear preference for models that remain accountable to physics.

Selected Papers

This digest features 40 selected papers from 951 papers analyzed.

1. Fast and effective classification of landslide microseismicity: a machine learning perspective

Source: Engineering Geology Type: Detection and Monitoring Geohazard Type: Landslides Relevance: 10/10

Core Problem: Reliable near-real-time classification of landslide-related microseismicity is difficult when monitoring is sparse and environmental variability distorts spectral signatures.

Key Innovation: The study compares supervised and unsupervised single-station workflows and shows that a CNN with epistemic-uncertainty estimation is more robust than clustering under non-stationary field conditions.

2. Effect of Phyllosilicate Enrichment on the Residual Shear Strength of Thrust Zone Landslide Soils in the Mid-Nepal Himalaya

Source: Geotech. & Geol. Eng. Type: Concepts & Mechanisms Geohazard Type: Landslides Relevance: 10/10

Core Problem: Residual shear strength in Himalayan roadside landslide soils cannot be explained adequately by clay content alone, obscuring how tectonic alteration weakens thrust-zone materials.

Key Innovation: Ring-shear testing and XRD support a new secondary-to-primary mineral index that links phyllosilicate enrichment directly to residual friction loss, offering a practical proxy for post-failure weakness.

3. Climate-impacted reliability analysis of earth slopes with spatially variable soils using active learning surrogate modeling

Source: Computers and Geotechnics Type: Risk Assessment Geohazard Type: Earth slopes Relevance: 10/10

Core Problem: Slope safety under long climatic records is hard to quantify because spatial soil variability, groundwater shifts, and extreme weather jointly alter reliability through time.

Key Innovation: A multi-level probabilistic framework couples FEM/LEM analyses with adaptive polynomial chaos-Kriging to update slope stability under 33 years of climate forcing and spatially variable soils.

4. Suspended sediment decline intensified riverbank erosion and collapse in the Yangtze river mainstream

Source: ISPRS J. Photogrammetry Type: Detection and Monitoring Geohazard Type: Riverbank erosion and collapse Relevance: 10/10

Core Problem: Long-term sediment decline in the Yangtze has been widely noted, but its reach-scale linkage to bank erosion and collapse remains poorly quantified.

Key Innovation: A Landsat-derived sediment concentration record shows how post-dam sediment starvation reorganized the main-stem gradient and strongly amplified erosion and bank-collapse frequency in downstream reaches.

5. Mechanics-informed machine learning: A gray-box framework for pore-pressure denoising

Source: Engineering Geology Type: Detection and Monitoring Geohazard Type: Ground deformation and subsidence risk Relevance: 9/10

Core Problem: Pore-pressure records from pumped aquifers are often dominated by environmental noise, obscuring the extraction-driven signals most relevant for subsidence and infrastructure risk.

Key Innovation: A gray-box framework couples poromechanics and machine learning to recover physically meaningful pore-pressure signals from noisy monitoring data and outperform purely statistical denoising strategies.

6. Integrated numerical-microseismic approach to unraveling the incubation mechanisms of strain burst and fault-slip rockburst

Source: TUST Type: Early Warning Geohazard Type: Rockburst Relevance: 9/10

Core Problem: Deep tunnel rockbursts remain difficult to forecast because strain bursts and fault-slip bursts exhibit different precursory patterns and energy-release pathways.

Key Innovation: Calibrated numerical modelling and moment-tensor-informed microseismic monitoring are integrated to distinguish the incubation sequences of tensile strain bursts versus structurally controlled fault-slip rockbursts.

7. Vertical-force-dominated mechanisms and failure processes in mining-induced seismicity via calibrated moment tensor inversion

Source: Intl. J. Rock Mech. & Mining Type: Concepts & Mechanisms Geohazard Type: Mining-induced seismicity Relevance: 9/10

Core Problem: Mining seismicity is often misinterpreted through shear-fault analogies, masking the true overburden failure mechanisms that control dangerous energy release.

Key Innovation: A physically constrained moment-tensor inversion reveals dominant vertical-force components and supports a key-stratum-controlled seismogenic model centered on tensile opening and compressive closure rather than fault slip.

8. Seismic response of shallow-buried composite lining tunnels in saturated soil: A theoretical analysis model and shaking table tests

Source: TUST Type: Hazard Modelling Geohazard Type: Tunnel seismic response Relevance: 9/10

Core Problem: Composite-lined tunnels in saturated strata lack a validated analytical description of seismic loading that accounts for pore-pressure effects and multilayer lining behavior.

Key Innovation: A Biot-based analytical model combined with shaking-table tests quantifies how inner-lining stiffness and thickness redistribute pore pressure and lining stress in shallow saturated tunnels.

9. Hydraulic fracturing-based static rock burst prevention in thick and hard siltstone roofs: A multi-scale numerical and field investigation

Source: TUST Type: Mitigation Geohazard Type: Rockburst Relevance: 9/10

Core Problem: Static-load rockburst in deep coal roadways persists because thick, hard roofs store elastic energy regionally rather than through isolated local triggers.

Key Innovation: Underground ultra-long segmented hydraulic fracturing is shown to reduce roof stiffness, coal-pillar stress, and microseismic activity, establishing a regional pressure-relief strategy for static rockburst control.

10. Seismic effects on onshore wind turbines: A critical engineering perspective for resilience and disaster risk reduction

Source: IJDRR Type: Risk Assessment Geohazard Type: Earthquakes, infrastructure resilience Relevance: 8/10

Core Problem: Wind turbines are increasingly deployed in seismically active regions, yet design practice still underestimates the interaction of wind, seismic loading, and soil response.

Key Innovation: This synthesis shows how near-fault pulses, operating state, and soft-soil interaction can shift demand beyond wind-governed design assumptions, making unified multi-hazard criteria necessary.

11. Study on spatial variation characteristics of seismic motion in near-fault zones based on ground motion simulation and probabilistic seismic hazard analysis: A case study of the Nankou-Sunhe Fault

Source: Soil Dyn. & Earthquake Eng. Type: Hazard Modelling Geohazard Type: Earthquakes Relevance: 8/10

Core Problem: Near-fault design still relies heavily on empirical adjustments despite strong segment-scale spatial heterogeneity in shaking intensity.

Key Innovation: Stochastic finite-fault simulations and PSHA map segmented high-PGA belts and validate a 5 km near-fault threshold, offering more physically grounded spatial design inputs.

12. Field-Scale Study of Rock Scour in an Unlined Spillway

Source: ASCE J. Geotech. Geoenviron. Type: Concepts & Mechanisms Geohazard Type: Rock scour, dam safety Relevance: 8/10

Core Problem: Rock scour thresholds derived from scaled laboratory studies remain difficult to transfer confidently to full spillway conditions.

Key Innovation: Instrumented artificial blocks installed in an operating spillway provide prototype pressure and displacement data during plucking, helping anchor three-dimensional block-theory scour assessments to field reality.

13. Flood and heavy metal risks from wastewater site in Sohag Governorate, Egypt: integrating hydrological modeling and mapping

Source: Env. Earth Sciences Type: Risk Assessment Geohazard Type: Floods, contaminant transport Relevance: 8/10

Core Problem: Flood-prone wastewater disposal sites pose coupled inundation and contaminant-transfer hazards that are rarely evaluated together.

Key Innovation: Hydrological simulations and soil geochemistry are combined to show how floodwater can mobilize heavy metals from disposal areas into canals and the Nile, clarifying compound environmental risk.

14. An integrated framework for mapping agricultural water impoundments using Sentinel −2 and GEE in Northwest China

Source: Journal of Hydrology Type: Detection and Monitoring Geohazard Type: Water scarcity and irrigation risk Relevance: 8/10

Core Problem: Small agricultural impoundments are difficult to inventory consistently, yet they are increasingly important to water security in arid regions.

Key Innovation: A persistence-plus-morphometry mapping framework delivers the first high-resolution multi-year inventory of agricultural water impoundments across the Hexi Corridor.

15. Physics-informed deep learning for hillslope subsurface flow prediction: integrating soil moisture deficit constraints

Source: Journal of Hydrology Type: Hazard Modelling Geohazard Type: Runoff generation, flood response Relevance: 8/10

Core Problem: Hillslope subsurface stormflow is threshold-controlled and intermittent, making purely data-driven models physically inconsistent under changing moisture states.

Key Innovation: A physics-informed LSTM uses soil-moisture-deficit constraints to gate rainfall response, improving both predictive skill and hydrological consistency in threshold-driven runoff generation.

16. Exploring a process-aware spatiotemporal graph-based surrogate for integrated urban drainage simulation

Source: Journal of Hydrology Type: Hazard Modelling Geohazard Type: Urban flooding Relevance: 8/10

Core Problem: Urban drainage models remain too computationally expensive for many real-time flood-control applications, while existing surrogates often lose process realism.

Key Innovation: A process-aware graph surrogate embeds rainfall-runoff-routing structure and control logic, achieving high-fidelity water-depth and inflow prediction under both regulated and extreme rainfall conditions.

17. Effects of Dissolution Regimes on Flow Channelization and Solute Transport in 3D Fracture Networks: Insights From Graph‐Based Reactive Transport Modeling

Source: JGR: Earth Surface Type: Concepts & Mechanisms Geohazard Type: Fractured-rock flow and transport hazards Relevance: 8/10

Core Problem: The way mineral dissolution reorganizes flow paths in realistic fracture networks remains difficult to quantify at a scale relevant to subsurface engineering and remediation.

Key Innovation: A graph-based reactive transport model isolates dissolution-driven transitions from mild channelization to wormholing, clarifying how initial heterogeneity, advection, and reaction kinetics restructure transport pathways.

18. Dynamic Models of Magma Storage Within a Damaging, Strain‐Softening Crust and Their Application to Sierra‐Negra's Pre‐Eruptive Inflation Pattern

Source: JGR: Earth Surface Type: Concepts & Mechanisms Geohazard Type: Volcanic unrest Relevance: 8/10

Core Problem: Pre-eruptive inflation transients are often attributed to ad hoc changes in magma supply, leaving the mechanical feedbacks of crustal damage underexplored.

Key Innovation: A coupled magma-transport and strain-softening crust model reproduces Sierra Negra's transient uplift sequence by showing how damage-induced mechanical weakening modulates shallow storage dynamics.

19. Dynamic Earthquake Source Inversion With Generative Adversarial Network Priors

Source: JGR: Earth Surface Type: Hazard Modelling Geohazard Type: Earthquakes Relevance: 8/10

Core Problem: Large dynamic earthquake source inversion remains computationally prohibitive because high-dimensional Bayesian sampling converges too slowly.

Key Innovation: A staged inversion framework transfers prior information between temporal-spatial scales using Wasserstein GANs, enabling more stable dynamic rupture inference for larger earthquakes.

20. New Constraints on the Calabrian Arc (Central Mediterranean) Geodynamics: High‐Resolution Imaging of Lithospheric and Upper Mantle Discontinuities From P and S Receiver Functions

Source: GRL Type: Concepts & Mechanisms Geohazard Type: Earthquakes, subduction dynamics Relevance: 8/10

Core Problem: The relationship between shallow crustal structure and deeper slab tearing in the Calabrian Arc has remained poorly resolved.

Key Innovation: Joint inversion of dense P and S receiver-function data images Moho doubling, lithospheric tears, and mantle heterogeneity at high resolution, refining the structural context for regional deformation and seismicity.

21. Near‐Bed Flow Turbulence Beneath Ice Cover Under Varying Hydrological Conditions: A 9‐Year Field Measurement‐Based Analysis From a Meander Bend

Source: Water Resources Research Type: Detection and Monitoring Geohazard Type: Ice-covered river hazards Relevance: 7/10

Core Problem: Under-ice turbulence remains poorly constrained despite its importance for sediment transport, channel change, and winter flood processes.

Key Innovation: Nine years of field data reveal how water-column height, ice thickness, discharge, and bend geometry control near-bed turbulence beneath ice cover.

22. Influence of vegetation change on drip water dynamics and geochemical proxies in a shallow karst cave system

Source: Journal of Hydrology Type: Detection and Monitoring Geohazard Type: Karst hydrology, flood infiltration Relevance: 7/10

Core Problem: Drip-water geochemistry is widely used as a climate proxy, but vegetation-driven infiltration changes can strongly alter signal transmission through shallow karst systems.

Key Innovation: Seven years of monitoring show that changing vegetation and reservoir pathways can shift drip-water isotopic behavior from long-term rainfall integration toward heavy-rainfall dominance.

23. Unsupervised machine learning applied to Sentinel-2 time series for monitoring coastal lagoon water quality

Source: Journal of Hydrology Type: Detection and Monitoring Geohazard Type: Coastal water-quality hazards Relevance: 7/10

Core Problem: Long coastal-lagoon time series are difficult to parse for degradation hot spots and extreme-event anomalies without scalable automated workflows.

Key Innovation: Unsupervised clustering and STL decomposition are fused on Sentinel-2 records to identify ecologically distinct lagoon zones and reveal long-term turbidity and chlorophyll anomalies.

24. Effect of water salinity on the soil freezing-thawing characteristic curve of sandy soils

Source: Journal of Hydrology Type: Concepts & Mechanisms Geohazard Type: Frozen ground Relevance: 7/10

Core Problem: Freeze-thaw behavior in saline sandy soils remains difficult to represent because salinity, pore structure, and supercooling interact nonlinearly.

Key Innovation: Laboratory measurements show how increasing salinity shifts control from soil-matrix dominance toward salt-driven inhibition of freezing, refining the hysteresis structure of saline freeze-thaw curves.

25. Seasonal variation patterns and drivers of baseflow recession dynamics across Australian

Source: Journal of Hydrology Type: Detection and Monitoring Geohazard Type: Hydrological drought Relevance: 7/10

Core Problem: Temporal shifts in baseflow recession remain poorly constrained, limiting drought forecasting and low-flow management.

Key Innovation: Event-scale recession analysis across 596 Australian catchments shows that seasonal variation dominates recession behavior and is largely controlled by vegetation, temperature, and evaporative demand.

26. Large lake buffers regional extreme heat and surface energy under the 2022 record-breaking heatwave in China

Source: Journal of Hydrology Type: Hazard Modelling Geohazard Type: Extreme heat Relevance: 7/10

Core Problem: The feedback role of large lakes during severe heatwaves remains underrepresented in regional hazard analysis.

Key Innovation: WRF experiments with and without Poyang Lake show that the lake acts as a phase-selective thermal buffer, damping daytime heat while modulating nighttime warming through latent heat flux and storage dynamics.

27. Extreme rainfall intensifies nitrogen responses through baseflow in arid and semi-arid river basins

Source: Journal of Hydrology Type: Risk Assessment Geohazard Type: Extreme rainfall, water-quality hazards Relevance: 7/10

Core Problem: Pollution responses to extreme rainfall are often attributed to quickflow pulses, leaving the role of baseflow under event-scale extremes underexplored.

Key Innovation: Hydrological modelling and interpretable machine learning show that baseflow can switch from dilution to nitrogen mobilization once a threshold is crossed during extreme rainfall in the Yellow River Basin.

28. Effect of supercritical CO2 on the anisotropic and seepage characteristics of shale

Source: Engineering Geology Type: Hazard Modelling Geohazard Type: Shale integrity and seepage risk Relevance: 7/10

Core Problem: Supercritical CO2 alters shale microstructure and anisotropy in ways that are crucial for sequestration and stimulation performance but remain insufficiently resolved.

Key Innovation: Experiments and digital-core analysis reveal a two-stage evolution from dissolution-driven connectivity increase to restructuring-driven local closure, with corresponding shifts in seepage anisotropy.

29. Hydro-mechanical and gas transport behaviour of crushed opalinus clay

Source: Engineering Geology Type: Concepts & Mechanisms Geohazard Type: Repository gas transport and sealing risk Relevance: 7/10

Core Problem: Crushed shale backfill behavior remains insufficiently characterized, especially the microstructural controls on gas dissipation and swelling in repository-like settings.

Key Innovation: An integrated campaign links pore-structure evolution, water retention, and gas breakthrough behavior in crushed Opalinus Clay, clarifying why persistent macropores can govern post-breakthrough dissipation.

30. Research on the seepage characteristics of fractured rock with different roughness during mining processes

Source: Bull. Eng. Geol. & Env. Type: Concepts & Mechanisms Geohazard Type: Seepage-induced mining hazards Relevance: 7/10

Core Problem: The coupled evolution of fracture roughness and seepage pathways under mining stress paths is not well enough understood for water-inrush prevention.

Key Innovation: Triaxial seepage tests and a dedicated path-search algorithm reveal how unloading temporarily boosts permeability before roughness-driven path reorganization causes exponential decline.

31. A Numerical Model for Considering Discontinuity Load Transfer Effect on Grouted Rockbolt with Complex Interface and Rockbolt Behavior

Source: Rock Mech. & Rock Eng. Type: Hazard Modelling Geohazard Type: Underground support failure Relevance: 7/10

Core Problem: Bond defects and nonlinear interface behavior can radically alter the pull-out response of grouted rockbolts, yet are poorly represented in practical design models.

Key Innovation: A defect-aware numerical model captures nonlinear slip, bolt plasticity, and snap-back behavior, clarifying how defect size and location should inform bolt selection and design.

32. Face stability analysis of deep tunnels crossing water-rich weak zone

Source: JRMGE Type: Hazard Modelling Geohazard Type: Tunnel face instability, mud-water inrush Relevance: 7/10

Core Problem: Limit support pressure at deep tunnel faces in water-rich weak zones is still uncertain because seepage-force estimation remains approximate in many models.

Key Innovation: An analytical framework solves hydraulic head and seepage forces directly from boundary conditions, producing a more accurate support-pressure estimate for inrush-prone tunnel faces.

33. A framework for automatic discontinuity trace extraction using multi-scale surface variation index and transfer-learning enhanced artificial neural network

Source: JRMGE Type: Detection and Monitoring Geohazard Type: Rock mass discontinuity mapping Relevance: 7/10

Core Problem: Rapid extraction of discontinuity traces from rock-mass point clouds remains too dependent on manual thresholds and site-specific tuning.

Key Innovation: A multi-scale surface-variation index and transfer-learning-enhanced ANN deliver robust threshold-free trace extraction and a high F1 score across rock-mass datasets.

34. Coupled FEM-DEM modeling of permeability evolution in rough fractured shale during shearing under varying confining pressures

Source: JRMGE Type: Concepts & Mechanisms Geohazard Type: Fractured-shale seepage Relevance: 7/10

Core Problem: Permeability evolution during shearing in rough fractured shale depends on dilation, debris production, and confining stress in ways that are difficult to observe directly.

Key Innovation: A coupled FEM-DEM framework reproduces how confining pressure suppresses dilation, enhances gouge generation, and shifts permeability evolution from transient increase to continuous decline.

35. A novel hydro-mechanical-chemical coupled model considering the transport of insoluble substances for acid stimulation

Source: Acta Geotechnica Type: Hazard Modelling Geohazard Type: Reservoir stimulation and flow-path instability Relevance: 7/10

Core Problem: Acid stimulation efficiency is controlled not only by dissolution but by stress coupling and insoluble byproducts that can alter wormhole evolution.

Key Innovation: A hydro-mechanical-chemical model introduces a porosity diffusion coefficient for insoluble-substance transport, showing how blockage can switch dissolution from wormholing to conical or face-dissolution modes.

36. Study on physical aperture and fluid flow behavior after dislocation of a rough fracture

Source: Intl. J. Rock Mech. & Mining Type: Concepts & Mechanisms Geohazard Type: Fractured-rock flow anisotropy Relevance: 7/10

Core Problem: Flow in dislocated rough fractures is difficult to quantify because aperture, roughness, and non-Darcy response evolve together with shear offset.

Key Innovation: Photogrammetry, 3D printing, and flow testing show how dislocation increases hydraulic aperture and anisotropy, while also benchmarking what printed replicas can and cannot reproduce.

37. Generalized interpolation phase field material point method for mixed-mode rock fracture

Source: Intl. J. Rock Mech. & Mining Type: Hazard Modelling Geohazard Type: Rock fracture Relevance: 7/10

Core Problem: Compression-shear mixed fracture under true triaxial stress remains challenging for phase-field formulations built mainly for tension-dominated cracking.

Key Innovation: An explicit PF-GIMP framework embeds a three-dimensional strength criterion into a material-point method, enabling stable simulation of mixed-mode rock fracture under multiaxial loading.

38. Impact of hydraulic fracturing combined with CO2 gas injection displacement on coalbed methane extraction

Source: Intl. J. Rock Mech. & Mining Type: Hazard Modelling Geohazard Type: Coal seam permeability and gas flow risk Relevance: 7/10

Core Problem: The coupled productivity, water-blocking, and sequestration consequences of combining hydraulic fracturing with CO2-enhanced coalbed methane extraction are not fully resolved.

Key Innovation: A fully coupled THMD model shows how fracturing and high-pressure CO2 injection jointly improve methane production and sequestration while mitigating long-term water blocking.

39. Effect of water content on small-strain dynamic properties of rock using free-free testing

Source: JRMGE Type: Detection and Monitoring Geohazard Type: Dynamic rock response Relevance: 7/10

Core Problem: Dynamic rock properties used in seismic and geotechnical analyses are often treated as fixed, even though water content can strongly alter wave speeds and damping.

Key Innovation: Low-cost free-free testing shows that rising water content can greatly increase Poisson's ratio and damping, establishing practical correction relationships for dynamic rock characterization.

40. Determination of failure envelope for submarine pipeline in spatially variable clay

Source: Acta Geotechnica Type: Risk Assessment Geohazard Type: Submarine pipeline instability Relevance: 7/10

Core Problem: Deterministic failure envelopes for pipelines in clay neglect spatial soil variability and can misrepresent design safety under combined loading.

Key Innovation: Random finite element analyses within a Monte Carlo framework derive probability-conditioned VH failure envelopes and show that code-style safety factors can be insufficient in random soils.