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

TerraMosaic Daily Digest: April 10, 2026

April 10, 2026
TerraMosaic Daily Digest

Daily Summary

This April 10, 2026 digest distills 34 selected papers from 915 analyzed records. The clearest pattern in the selected set is that the best papers identify specific structures that control later hazard expression. The most directly landslide-centered study reconstructs the flood-derived event layer implicated in the 2010 St. Jude landslide and shows that retrogressive failure in Champlain Sea sediments is partly a problem of inherited basin stratigraphy. The adjacent top papers make a similar move in different settings: subsidence in Taiwan is separated from crop intensity by multi-year SAR, Arctic ice-jam flooding is resolved through SWOT backwater profiles, and seismic bearing capacity on layered slopes changes once spatial soil variability is treated as a governing part of the system.

The rest of the selection broadens from those mechanistic cores into operational observation of hazard progression. Hourly wildfire spread, building-level earthquake damage proxies, Tibetan lake-ice thickness, Chinese lake-water storage, and Ethiopian groundwater drought are all analyzed as evolving trajectories rather than static end states. The strongest method papers also stay close to the physics they are trying to emulate: soil-moisture learning is decomposed into interpretable process terms, granular fault gouge is analyzed through contact-scale friction evolution, and inverse poroelastic calibration remains embedded in the governing equations. What makes the day coherent is not a single hazard type, but a repeated insistence on identifying the pathway, storage term, or inherited structure that determines why similar forcing produces different consequences.

Key Trends

The strongest papers today gain leverage by identifying the specific structure or pathway that turns forcing into consequence.

  • Inherited structure is emerging as a first-order hazard control: The strongest papers explain later failure and deformation by recovering buried event layers, conductive ascent pathways, groundwater-storage organization, and spatial soil variability that ordinary surface descriptions miss.
  • Hazard remote sensing is becoming both finer in time and sharper in object scale: This day’s selected papers move from static snapshots to hourly wildfire spread, individual-building earthquake damage, multi-year subsidence, ice-jam backwater profiles, and evolving lake and glacier states.
  • Storage, drainage, and transmission are central across hydroclimatic and geotechnical hazards: Groundwater memory, soil-moisture process partitioning, lake storage, crack-assisted infiltration, and pile drainage under liquefaction all determine whether stress is buffered or amplified.
  • Method papers matter here when they preserve physical meaning: The most useful AI and numerical advances in today’s set are the ones that keep process variables visible, whether in soil-moisture learning, slope-capacity surrogates, fault-gouge friction, or differentiable poroelastic inversion.

Selected Papers

This digest features 34 selected papers from 915 papers analyzed, spanning retrogressive landslides, subsidence, ice-jam floods, volcanic pathways, wildfire progression, earthquake damage, groundwater drought, cryosphere change, and underground geomechanics.

1. A catastrophic flood event layer in Champlain Sea sediments recorded at the 2010 St. Jude landslide (Québec, Canada): regional stratigraphic insights from piezocone data

Source: Can. Geotech. J. Type: Concepts & Mechanisms Geohazard Type: Retrogressive landslides Relevance: 9/10

Core Problem: Large retrogressive landslides in Champlain Sea sediments are still interpreted mostly from local profiles, leaving the basin-scale sedimentary controls on failure-surface localization poorly resolved.

Key Innovation: Piezocone evidence identifies a widespread event layer tied to a catastrophic outburst-flood deposit and shows that this inherited stratigraphy likely guides failure surfaces in major retrogressive landslides across the St. Lawrence Valley.

2. Interlinked Dynamics of Rice Cultivation Intensity and Ground Subsidence: Spatio‐Temporal Insights Into Water Resource Implications From Multi‐Year Spaceborne SAR Analysis

Source: Water Resources Research Type: Detection and Monitoring Geohazard Type: Ground subsidence Relevance: 9/10

Core Problem: Ground subsidence in irrigated agricultural regions is often attributed to cultivation intensity without sufficiently separating crop patterns from broader groundwater-management effects.

Key Innovation: An eight-year SAR analysis jointly tracks rice intensity and subsidence and shows that persistent agricultural hotspots are not simply co-located with subsiding zones, redirecting attention toward groundwater-well management rather than blanket farmland exclusion.

3. Backwater Effects From River Ice Jams Observed With SWOT Pixel Cloud Data

Source: GRL Type: Detection and Monitoring Geohazard Type: Ice-jam floods Relevance: 8/10

Core Problem: Ice jams can generate abrupt upstream flooding and dangerous downstream release waves, yet their hydraulic signatures have been difficult to observe continuously at river scale.

Key Innovation: SWOT pixel-cloud elevation profiles directly recover upstream backwater flattening caused by Arctic ice jams, demonstrating a new satellite route for quantifying ice-jam hydraulics and flood risk.

4. Magma and Volatile Pathways Beneath Sakurajima Volcano From Self‐Potential, Helium Isotopes, and Broadband Magnetotellurics

Source: GRL Type: Concepts & Mechanisms Geohazard Type: Volcanic hazards Relevance: 8/10

Core Problem: At persistently erupting volcanoes, magma and volatile ascent pathways are usually inferred from isolated datasets that do not resolve how deep reservoirs branch into shallow conduits.

Key Innovation: An integrated self-potential, helium-isotope, and magnetotelluric analysis reveals a long-lived conductive reservoir beneath Sakurajima and supports an edge-ascent model in which magma and volatiles rise along reservoir boundaries before shallow lateral diversion.

5. The Slip Distributions of the 1952 and 2025 Kamchatka Earthquakes From Tsunami Waveforms Recorded Around the Pacific Ocean

Source: GRL Type: Hazard Modelling Geohazard Type: Megathrust earthquakes, tsunamis Relevance: 8/10

Core Problem: The unexpected recurrence of a giant Kamchatka earthquake in 2025 raised immediate questions about whether it ruptured the same patch as the 1952 event or a mechanically distinct segment.

Key Innovation: Pacific-wide tsunami inversions show broadly similar slip patterns for the 1952 and 2025 Kamchatka earthquakes but with important differences in peak-slip depth and northern rupture extent, refining how repeated giant events may reuse a subduction interface.

6. Satellite-based analysis of hourly progression and driving factors of large U.S. wildfires

Source: International Journal of Applied Earth Observation and Geoinformation Type: Detection and Monitoring Geohazard Type: Wildfires Relevance: 8/10

Core Problem: Daily wildfire growth products miss the hourly spread bursts that matter most for suppression and evacuation timing.

Key Innovation: A GOES-based dataset for 294 large U.S. fires resolves hourly progression nationwide, identifies the fastest regions and events, and ranks wind speed as the dominant driver of rapid spread.

7. SAR-based individual-building damage identification and large-scale earthquake damage prediction

Source: International Journal of Applied Earth Observation and Geoinformation Type: Detection and Monitoring Geohazard Type: Earthquakes Relevance: 8/10

Core Problem: Post-earthquake response still lacks reliable products that identify structural damage at the scale of individual buildings across large urban regions.

Key Innovation: A coherence-based Sentinel-1 framework produces building-level damage proxy maps with strong accuracy after the 2023 Türkiye earthquakes and couples them to regional damage-density prediction when direct observations are delayed.

8. Deep learning for the static and seismic bearing capacity of shallow strip footings on multi-layered slopes considering spatial soil variability

Source: Computers and Geotechnics Type: Hazard Modelling Geohazard Type: Slope instability, seismic footing failure Relevance: 8/10

Core Problem: Bearing-capacity assessment on slopes rarely captures both spatially variable soil structure and seismic loading without prohibitive stochastic simulation cost.

Key Innovation: A physics-aligned deep learning surrogate trained on CS-FEM stochastic simulations reproduces static and seismic bearing capacity across layered random slopes while preserving non-stationary spatial effects.

9. Coupling Intuitive Physics Into Deep Learning for Soil Moisture Flow Processes Learning

Source: Water Resources Research Type: Hazard Modelling Geohazard Type: Soil moisture, drought and infiltration hazards Relevance: 7/10

Core Problem: Deep learning can reconstruct soil-moisture dynamics well, but its internal reasoning is usually too opaque to support process interpretation in hydrologic hazard contexts.

Key Innovation: The DPL-S framework decomposes soil-moisture evolution into physically legible sub-process tensors and retains competitive predictive skill while making gravity, matric potential, and scene dependence explicitly interpretable.

10. Space–time modeling of Net Primary Productivity before and after major earthquakes

Source: International Journal of Applied Earth Observation and Geoinformation Type: Risk Assessment Geohazard Type: Earthquakes, coseismic landslides Relevance: 7/10

Core Problem: Earthquake-triggered vegetation loss and carbon-sink disruption remain poorly quantified beyond isolated post-event case studies.

Key Innovation: A space-time attention framework reconstructs NPP disturbance and recovery across ten major earthquake regions and converts carbon-sink disruption into monetary loss estimates.

11. A first characterization of lake ice thickness on the Tibetan Plateau by leveraging satellite altimetry and ERA5 reanalysis

Source: International Journal of Applied Earth Observation and Geoinformation Type: Detection and Monitoring Geohazard Type: Cryosphere and lake hazards Relevance: 7/10

Core Problem: Lake ice thickness on the Tibetan Plateau has lacked consistent large-scale observation, limiting understanding of changing cryospheric storage and lake response.

Key Innovation: Multi-mission altimetry and reanalysis are combined to produce the first broad observational characterization of Tibetan Plateau lake ice thickness and show an overall thinning trend driven mainly by temperature.

12. Long-term deformation monitoring and potential building damage assessment in resource-based arid urban agglomerations

Source: International Journal of Applied Earth Observation and Geoinformation Type: Detection and Monitoring Geohazard Type: Land subsidence, urban damage Relevance: 7/10

Core Problem: County-scale surface deformation in arid resource-based urban belts is often described geodetically without a building-level translation into potential structural damage.

Key Innovation: Long-term ALOS and Sentinel InSAR across the HBOY urban agglomeration maps persistent subsidence growth and links deformation fields to building footprints to estimate differential damage intensity.

13. Characterization of groundwater storage dynamics and responses to meteorological stress and climate teleconnections across river basins and climate zones in Ethiopia

Source: Journal of Hydrology Type: Risk Assessment Geohazard Type: Groundwater drought Relevance: 7/10

Core Problem: Groundwater drought in Ethiopia remains weakly characterized because sparse in-situ monitoring has forced overreliance on precipitation indices.

Key Innovation: GRACE-based storage anomalies and a standardized groundwater drought index reveal strong basin-to-basin heterogeneity, long groundwater memory, and scale-dependent teleconnection effects across Ethiopian climate zones.

14. Hydro-meteorological drivers of spatiotemporal variability in lake water storage across China: Evidence from ICESat-2

Source: Journal of Hydrology Type: Detection and Monitoring Geohazard Type: Floods, droughts, lake storage hazards Relevance: 7/10

Core Problem: The drivers of seasonal lake-water-storage variability remain poorly separated across China’s contrasting lake regions.

Key Innovation: ICESat-2-based water levels and machine-learning gap filling resolve nationwide storage variability and show regionally distinct responses to compound meteorological and hydrological drought controls.

15. Multidimensional aspects of drought event evolution drive the spatiotemporal heterogeneity of vegetation photosynthetic responses

Source: International Journal of Applied Earth Observation and Geoinformation Type: Risk Assessment Geohazard Type: Droughts Relevance: 6/10

Core Problem: Vegetation carbon-response studies often compress drought into static severity descriptors and miss how event evolution and migration shape ecological damage.

Key Innovation: A three-dimensional clustering of drought lifecycles across China links distinct event trajectories to regionally different photosynthetic sensitivity thresholds and drought impacts.

16. Mapping burned areas by unsupervised change detection combining automatic multi-index driven label generation and ensemble learning

Source: International Journal of Applied Earth Observation and Geoinformation Type: Detection and Monitoring Geohazard Type: Wildfires Relevance: 6/10

Core Problem: Rapid burned-area mapping still struggles to avoid manual labels while remaining robust to threshold choice, spatial context, and historical-burn interference.

Key Innovation: An unsupervised Sentinel-2 framework generates its own labels from multiple burn indices and neighborhood rules, then uses ensemble learning and spatial statistics to improve rapid burned-area delineation.

17. Uncertainty-aware few-shot snow depth estimation in complex terrains via generative-Bayesian deep learning

Source: International Journal of Applied Earth Observation and Geoinformation Type: Hazard Modelling Geohazard Type: Snow hazards Relevance: 6/10

Core Problem: Snow-depth estimation in complex terrain is constrained by sparse ground measurements and few-shot training conditions that undermine generalization.

Key Innovation: A generative-Bayesian framework augments underrepresented snow regimes and yields uncertainty-aware snow-depth estimation with strong skill under data-limited conditions.

18. Decadal glacier flow acceleration caused by upper ocean warming in the Antarctic Peninsula

Source: International Journal of Applied Earth Observation and Geoinformation Type: Detection and Monitoring Geohazard Type: Cryosphere hazards, glacier change Relevance: 6/10

Core Problem: Decadal glacier acceleration along the Antarctic Peninsula is still difficult to attribute cleanly between atmospheric and oceanic forcing.

Key Innovation: Remote-sensing analysis ties sustained glacier-flow acceleration to upper-ocean warming, sharpening the mechanism linking marine forcing to long-term cryospheric instability.

19. Evaluating Pléiades Neo capabilities for deriving rock glacier velocity

Source: International Journal of Applied Earth Observation and Geoinformation Type: Detection and Monitoring Geohazard Type: Rock glaciers Relevance: 6/10

Core Problem: Rock-glacier flow is often monitored at coarse temporal or spatial resolution, limiting annual-scale detection of subtle kinematic change in remote terrain.

Key Innovation: Repeated Pléiades Neo acquisitions in the Dry Andes produce high-detail annual rock-glacier velocity fields and show that high-resolution commercial imagery can now support routine remote kinematic monitoring.

20. A damage-based numerical approach for optimizing the arrangement of articulated sections in an active-fault crossing linear underground structure

Source: Computers and Geotechnics Type: Hazard Modelling Geohazard Type: Fault rupture hazards Relevance: 6/10

Core Problem: Articulated design is widely used for fault-crossing underground structures, but arrangement geometry has seldom been optimized systematically against expected damage length.

Key Innovation: A COMSOL-MATLAB damage framework evaluates thousands of articulated configurations and shows how section number, width, and spacing materially control longitudinal fault damage.

21. Numerical insights into rate-and-state frictional control of quasi-periodic stick–slip in granular fault gouge

Source: Computers and Geotechnics Type: Concepts & Mechanisms Geohazard Type: Earthquakes Relevance: 6/10

Core Problem: The grain-scale route from rate-and-state friction to stick-slip behavior in granular fault gouge remains only partly resolved.

Key Innovation: Discrete-element simulations with rate-and-state friction reveal staged micro-friction evolution, rupture-front weakening, and a scaling link between co-seismic slip, stress drop, and fracture energy.

22. Seismic performance of permeable pipe piles in liquefiable site using shaking table tests

Source: Soil Dyn. & Earthquake Eng. Type: Mitigation Geohazard Type: Liquefaction hazards Relevance: 6/10

Core Problem: Conventional pipe piles do not directly dissipate excess pore pressure during shaking, leaving liquefaction-prone foundations vulnerable to rapid strength loss.

Key Innovation: Large-scale shaking-table tests show that permeable pipe piles with drainage holes sharply reduce excess pore pressure, deformation, and pile-head displacement in liquefiable foundations.

23. A comprehensive finite element model for nonlinear seismic response of rockfill dam-reservoir-foundation system

Source: Soil Dyn. & Earthquake Eng. Type: Hazard Modelling Geohazard Type: Dam safety, earthquakes Relevance: 6/10

Core Problem: Rockfill-dam seismic assessments often simplify reservoir interaction, slab behavior, or foundation damping and therefore miss key system couplings during strong shaking.

Key Innovation: A fully coupled finite-element model of the dam-reservoir-foundation system reproduces the observed Zipingpu Dam response more faithfully than standard simplified approaches.

24. Strain arch-bottom rockbursts in deep D-shaped tunnels under high horizontal principal stress

Source: JRMGE Type: Concepts & Mechanisms Geohazard Type: Rockbursts Relevance: 6/10

Core Problem: Arch-bottom rockbursts in deep tunnels remain poorly reproduced experimentally, leaving their triggering stress path and staged morphology uncertain.

Key Innovation: True-triaxial visualization tests reproduce arch-bottom rockbursts, identify horizontal principal-stress adjustment as the key trigger, and resolve the staged evolution toward a deep V-shaped failure pit.

25. Crack-hydro-deformation analysis of expansive soil: implications on heave–subsidence response

Source: Can. Geotech. J. Type: Concepts & Mechanisms Geohazard Type: Expansive-soil hazards Relevance: 6/10

Core Problem: Desiccation cracks in expansive soils are known to alter infiltration, but their role in seasonal heave-subsidence damage along pavement margins has remained weakly quantified.

Key Innovation: A dual-permeability unsaturated model shows that crack networks strongly amplify moisture ingress and differential vertical deformation, especially as crack depth increases.

26. EA-ERT: a new ensemble approach to convert time-lapse ERT data to soil water content

Source: Journal of Hydrology Type: Detection and Monitoring Geohazard Type: Soil moisture, slope and drought monitoring Relevance: 5/10

Core Problem: Converting time-lapse electrical resistivity into spatial water-content fields remains highly sensitive to inversion choices and uncertainty structure.

Key Innovation: An ensemble ERT workflow calibrates multiple inversion models against field measurements, delivers robust water-content retrievals, and explicitly maps uncertainty hotspots in the final subsurface estimate.

27. Stochastic modeling of two-dimensional transient unconfined groundwater flow in terms of its time–space evolutionary probability distribution

Source: Journal of Hydrology Type: Hazard Modelling Geohazard Type: Groundwater hazards Relevance: 5/10

Core Problem: Groundwater-flow prediction at watershed scale remains sensitive to uncertain conductivity, source terms, and boundaries that deterministic models cannot represent well.

Key Innovation: A Lagrangian-Eulerian extension of the Fokker-Planck framework captures the evolving probability distribution of transient unconfined flow and matches Monte Carlo behavior with far lower cost.

28. A generalized enhanced snow/ice index based on Landsat imagery for accurate snow/ice mapping in mountainous areas

Source: Journal of Hydrology Type: Detection and Monitoring Geohazard Type: Snow, glacier and glacial-lake hazards Relevance: 5/10

Core Problem: Snow and ice mapping in mountainous terrain still suffers from vegetation mixing and snow-water confusion that reduce the reliability of standard indices.

Key Innovation: The generalized enhanced snow/ice index improves discrimination across seasonal snow, glaciers, and glacial lakes and transfers well across major mountain regions.

29. Analytical solution for seismic response of fault-adjacent deep tunnel to P1 waves with fluid-solid coupling effects

Source: Soil Dyn. & Earthquake Eng. Type: Hazard Modelling Geohazard Type: Tunnel earthquake hazards Relevance: 5/10

Core Problem: Deep tunnels near faults are difficult to assess analytically once poroelastic coupling and geological interfaces are both introduced.

Key Innovation: A Biot-theory analytical solution derives the P1-wave response of fault-adjacent deep tunnels in saturated media and clarifies how incidence, permeability, and porosity shape seismic demand.

30. Nondestructive integrity evaluation of soil nails using OFDR-based heated optical fibre

Source: JRMGE Type: Detection and Monitoring Geohazard Type: Slope stabilization infrastructure Relevance: 5/10

Core Problem: Soil-nail defects are difficult to identify non-destructively once installed, especially when defect length and grouting quality must both be inferred.

Key Innovation: Heated optical-fiber sensing combined with empirical mode decomposition resolves defect zones and estimates defect height and grouting compactness when thermal contrast is present.

31. Development of new relationships for estimating the in-situ stress in the Canadian Shield based on geological conditions

Source: JRMGE Type: Methods Geohazard Type: Deep geotechnical hazards Relevance: 5/10

Core Problem: Stress-depth relations for the Canadian Shield remain uncertain when geological diversity is flattened into arbitrary depth bins.

Key Innovation: Geologically grouped regression relations recover more representative in-situ stress patterns and provide a stronger basis for deep rock-mass stability assessment.

32. In-Poro: Differentiable Inverse Analysis Framework for Poroelasticity

Source: ASCE J. Geotech. Geoenviron. Type: Methods Geohazard Type: Poroelastic geohazards Relevance: 5/10

Core Problem: Spatial calibration of poroelastic material properties from displacement and pore-pressure data remains difficult with traditional inverse workflows limited to a few constants.

Key Innovation: A differentiable solver coupled to physics-constrained neural networks infers heterogeneous poroelastic property fields with high accuracy even under sparse and noisy observations.

33. Harvesting AlphaEarth: Benchmarking the geospatial foundation model for agricultural downstream tasks

Source: International Journal of Applied Earth Observation and Geoinformation Type: Methods Geohazard Type: Remote sensing foundation models Relevance: 5/10

Core Problem: AlphaEarth embeddings are promising, but their downstream value outside the original benchmark suite remains under-tested in realistic Earth-observation workflows.

Key Innovation: A dedicated evaluation across agricultural tasks shows where ready-made geospatial foundation embeddings are competitive with purpose-built remote-sensing models and where local data still dominate.

34. Predicting gas–water permeability in hydrate-bearing sediments considering hydrate morphology

Source: JRMGE Type: Hazard Modelling Geohazard Type: Gas hydrate geohazards Relevance: 4/10

Core Problem: Permeability evolution in hydrate-bearing sediments is difficult to predict because hydrate saturation, confining pressure, and pore habit jointly alter connectivity.

Key Innovation: Experiments and lattice-Boltzmann simulations yield a morphology-weighted permeability model that tracks the transition from grain-coating to pore-filling hydrate behavior.