TerraMosaic Daily Digest: June 12, 2026
Daily Summary
The June 12 issue is led by methods that make slope processes more explicit rather than more empirical. A differentiable simulator for rainfall-induced landslide motion embeds hydrological diffusion and viscoplastic flow inside the learning loop, allowing sparse satellite and rainfall records to identify physically admissible parameters. ASA-DETR addresses a different bottleneck in landslide work: small, texture-confounded failures in optical imagery. Its frequency-aware feature learning and sparse attention are paired with a new multi-sensor landslide detection dataset, placing object detection closer to operational geomorphic mapping.
The strongest geohazard papers treat failure as time-dependent deformation under changing hydraulic boundary conditions. Riverbank collapse is recast as a cascade from static susceptibility to InSAR-observed pre-failure motion; slow drawdown slope reliability is governed by a critical pool level where pore-pressure destabilization and external hydrostatic support trade off; and karst-slope experiments show how bedrock fissures partition repeated extreme rainfall between surface runoff, belowground flow, and soil loss. Geotechnical papers extend the same logic into engineered ground: helical soil nails, MICP-treated liquefiable sand, shallow tunnels under slope terrain, layered tunnel rock, and water-rich shield undercrossing are evaluated through mechanisms that couple construction, reinforcement, groundwater, and material heterogeneity.
A second thread links hazards to monitoring variables that can be observed before failure or impact. Very-low-frequency earthquake signals preceding a strike-slip earthquake provide rare evidence for a slow-to-fast rupture transition. Drone-based river-velocity inversion, remotely sensed surface-water storage, tree-ring flood reconstructions, Arctic shelf optical changes caused by thaw and coastal erosion, and alluvial anabranching models all translate geomorphic state into measurable hydrologic or sedimentary constraints. The remaining papers broaden the same measurement-first view to seismic strengthening, soil freezing, precipitation-tail risk, flood-prone informal settlements, coastal barriers, wave-attenuating slopes, and marine green-tide disasters.
Key Trends
Five movements define the issue: physics-constrained landslide learning, hydraulic boundary-condition analysis, system-scale geotechnical mitigation, precursor-based monitoring, and coupled cold-region/coastal/riverine geomorphology.
- Physics is moving inside landslide learning systems: The differentiable landslide simulator, ASA-DETR, and riverbank-collapse framework show a shift from black-box prediction toward models that encode flow laws, frequency-domain terrain texture, deformation history, or explainable causal structure.
- Hydraulic boundary conditions are becoming central to slope hazard analysis: Slow drawdown reliability, karst rainfall flume tests, shallow tunnel stress analysis, and water-rich shield undercrossing all identify instability through changing pore pressure, fissure flow, external water level, and excavation-induced stress redistribution.
- Geotechnical mitigation is being evaluated at system scale: Grouted-helical soil nails, MICP treatment extent, Himalayan masonry strengthening, deep underground self-recovering layers, and tunnel rock-support models test whether local interventions change the response of the whole slope, soil profile, structure, or excavation system.
- Hazard monitoring is shifting toward observable precursors and operational proxies: InSAR bank deformation, drone river velocity, surface-water storage, tree-ring flood reconstructions, hydroclimatic tail metrics, and VLFE precursors convert sparse observations into early warning or long-period risk constraints.
- Cold-region, coastal, and riverine processes are treated as coupled geomorphic systems: Soil freezing, Arctic coastal material export, barrier-island stratigraphy, alluvial anabranching, hollow-square vegetated slopes, karst depressions, and bridge-induced channel adjustment connect climate forcing, sediment routing, and infrastructure exposure.
Selected Papers
The selected papers emphasize physically constrained landslide simulation, remote-sensing landslide detection, riverbank collapse, drawdown-sensitive slope reliability, karst runoff and soil loss, liquefaction mitigation, earthquake initiation, river and coastal morphodynamics, and tunnel-ground interaction. This issue contains 33 selected papers from 1594 papers analyzed.
1. Physics-embedded differentiable simulation of rainfall-induced landslide motion
Core Problem: Slow-moving landslides are difficult to predict from sparse satellite and rainfall observations because purely empirical learning can violate hydromechanical constraints.
Key Innovation: Embeds hydrological diffusion and viscoplastic flow in a differentiable computational graph so landslide parameters can be inferred by automatic differentiation while predictions obey governing laws.
2. ASA-DETR: Adaptive sparse attention enhanced RT-DETR for remote sensing landslide detection with multi-scale frequency-aware feature learning
Core Problem: Complex landslide textures, scale variation, and small targets limit detection from optical imagery.
Key Innovation: Improves RT-DETR with a lightweight backbone, spatial-frequency feature learning, adaptive multi-scale attention, and the RSLD-2K dataset of 2,299 images and 6,545 landslide annotations.
3. Dynamic risk assessment of riverbank collapse: A cascading framework integrating explainable AI and InSAR
Core Problem: Static riverbank susceptibility maps cannot capture pre-failure deformation that controls near-term collapse risk.
Key Innovation: Combines an explainable stacking susceptibility model with InSAR deformation to build a cascading assessment from baseline risk to dynamic hazard.
4. Reliability analysis of slopes under slow drawdown conditions
Core Problem: Slow drawdown can produce maximum failure probability at a critical pool level rather than at the lowest water level.
Key Innovation: Uses random finite elements to quantify how shear strength, unit weight, slope height, depth ratio, internal pore pressure, and external hydrostatic pressure jointly define the critical drawdown state.
5. Effects of simulated successive extreme rainfall events on surface and belowground runoff and soil loss on flumes that mimic soil-mantled karst slopes
Core Problem: Repeated extreme rainfall on fissured karst slopes partitions water between surface and belowground pathways in ways that control erosion but remain poorly measured.
Key Innovation: Uses flumes representing 25-degree soil-mantled karst slopes with and without bedrock fissures to quantify runoff and soil-loss responses to three 100 mm h-1 rainfall events.
6. A generalized finite difference solution for stress and displacement around a shallow tunnel under slope terrain
Core Problem: Shallow tunnels beneath slopes experience asymmetric topographic loading that can threaten surrounding rock and slope stability.
Key Innovation: Develops a meshless generalized finite-difference solution for stress and displacement around shallow tunnels under slope terrain.
7. Development and Performance Evaluation of Grouted-Helical Soil Nails: An Experimental Study
Core Problem: Helical soil nails are easy to install but rotary installation can disturb surrounding soil and reduce pullout resistance.
Key Innovation: Tests a grouted-helical soil nail system that injects grout during installation to improve soil confinement and pullout performance.
8. Centrifuge Model Tests to Investigate the Effect of Microbially Induced Calcite Precipitation Treatment Extent within a Loose Liquefiable Sand Profile
Core Problem: The effect of partial MICP improvement depth on system-scale liquefaction performance remains uncertain.
Key Innovation: Uses centrifuge shaking tests on loose sand profiles with treatment depths from 37% to 100% of the liquefiable layer to evaluate pore-pressure generation, acceleration response, and liquefaction triggering.
9. Observational Evidence of a Very-Low-Frequency Earthquake (Mw 3.8) Leading to an Earthquake (Mw 4.2): Minto Flats Strike-Slip Fault Zone, Central Alaska
Core Problem: Immediate precursory observations before earthquake rupture are rare and controversial.
Key Innovation: Documents 43 seconds of high-frequency precursory signals including an Mw 3.8 very-low-frequency earthquake followed by an Mw 4.2 earthquake with a matching focal mechanism.
10. Subduction Parameters Controlling the Occurrence of Shallow and Deep Slow-Slip Events (SSEs)
Core Problem: The structural and kinematic controls on shallow and deep slow-slip events remain nonlinear and difficult to separate across subduction zones.
Key Innovation: Uses decision-tree machine learning to relate SSE occurrence to slab dip, convergence rate, slab roughness, lithostatic pressure, and along-interface heterogeneity.
11. Optimization of the cover and management factor (C) in the RUSLE model based on near soil surface characteristics for typical grasslands on the Chinese loess plateau
Core Problem: RUSLE C-factor estimation often omits near-surface traits such as litter, biological crusts, roots, and canopy cover.
Key Innovation: Quantifies how near-soil-surface characteristics of typical grasslands modify soil-erosion protection and improve C-factor estimation.
12. Effect of pasture restoration on runoff, soil loss, and sediment enrichment in a karst basin of the Brazilian savannah
Core Problem: Degraded tropical karst pastures can generate soil loss and downstream sedimentation, but restoration effects are poorly quantified.
Key Innovation: Measures runoff, soil loss, sediment enrichment, and restoration performance using field collectors in a Brazilian savannah karst basin.
13. Assessment of embedded risk in precipitation tails over India through concentration profiles: a multi-model assessment from CMIP6 experiments
Core Problem: Tail heaviness of daily precipitation does not directly express embedded risk across India's climate zones.
Key Innovation: Condenses concentration-profile information into tail-risk metrics across 4,801 grids and 16 CMIP6 models for historical and future periods.
14. Geomorphological and hydrological controls on shallow karst depressions (dayas) on a planar carbonate platform: the Nullarbor Plain, Australia
Core Problem: Shallow karst depressions in arid carbonate platforms are widespread but poorly constrained as hydrological storage features during extreme floods.
Key Innovation: Combines semi-automatic GIS mapping, morphometry, sediment analysis, and a March 2024 flood-retention assessment across 11 Nullarbor Plain study areas.
15. Shaking table test on seismic resilience improvement of a deep underground complex structure through Self-Recovering Layer
Core Problem: Deep underground complex structures can concentrate seismic demand and develop differential deformation at section transitions.
Key Innovation: Tests a Self-Recovering Layer in 1/40-scale shaking-table experiments to evaluate demand reduction and post-earthquake recoverability in soft soil.
16. Cutting performance and fragmentation behavior of rock under confining stress: role of heterogeneity
Core Problem: Rock cutting performance is controlled by microstructural heterogeneity and in-situ stress but is often oversimplified in predictive models.
Key Innovation: Combines a grain-based rock-cutting model with moment-tensor analysis to quantify fragmentation mechanisms under confining stress.
17. Grain Size and Fluid Effects on the Frictional Behavior of Simulated Carbonate Fault Gouge
Core Problem: The combined effects of grain-size distribution and pore fluid on carbonate fault-gouge friction remain poorly resolved.
Key Innovation: Uses rotary-shear slide-hold-slide experiments under dry, water, and geothermal brine conditions to constrain healing, creep, and frictional response.
18. Enhanced Solar Radiation Attenuation in the Arctic Shelf Driven by Warming-Induced Terrestrial Inputs
Core Problem: Arctic warming accelerates permafrost thaw and coastal erosion, changing material transport to shelf seas.
Key Innovation: Uses MODIS observations from 2002-2023 to identify increased diffuse attenuation in Arctic shelves linked to warming-driven runoff and coastal material export.
19. Remotely Sensed Surface Water Storage Shows Distinct Patterns From SWAT-Simulated Data
Core Problem: Hydrologic models often represent surface-water storage from static topography and calibrate mainly to discharge, missing dynamic storage behavior.
Key Innovation: Develops a Sentinel-1/2 remote-sensing storage dataset for the Upper Mississippi River Basin and compares it with SWAT simulations.
20. Modeling Length Adjustment in Alluvial Anabranching Units
Core Problem: Anabranching river units tend toward regular lengths, yet the mechanisms controlling elongation, contraction, and persistence are not well established.
Key Innovation: Formulates a dynamic model that balances base-flow elongation and higher-flow shortening to explain length adjustment in alluvial anabranching units.
21. Insights into greening the grey: Hydrodynamic analysis of wave attenuation on hollow-square slopes with and without vegetation
Core Problem: Hybrid vegetated coastal defenses need quantitative evidence for how blocks and plants dissipate wave energy.
Key Innovation: Combines flume experiments and SPH simulations to compare smooth slopes, hollow-square blocks, and plant-block systems.
22. The Apalachicola Barrier Island complex: a benchmark for MIS 5e (125 ka) sea-level oscillations?
Core Problem: Barrier islands preserve storm, sea-level, and shoreline-change histories that constrain coastal hazard baselines.
Key Innovation: Uses GPR, LiDAR, OSL ages, and paleogeographic reconstruction to resolve Pleistocene evolution of the Apalachicola Barrier Island complex.
23. Reframing Soil Freezing as an Unsaturated Process
Core Problem: Many soil-freezing models assume saturation, creating inconsistent hydraulic and thermal properties for partially frozen soils.
Key Innovation: Develops a fully coupled thermohydraulic model grounded in unsaturated soil mechanics and introduces a soil-freezing characteristic surface.
24. Lightweight Deep Learning for Real-Time River Surface Velocity Estimation From Low-Altitude Drone Remote Sensing
Core Problem: Field river-velocity monitoring needs both accurate water segmentation and low-latency inference on embedded drone platforms.
Key Innovation: Introduces AquaEdge, a lightweight framework that jointly segments water and estimates surface flow velocity from low-altitude drone imagery.
25. Affordable seismic strengthening of stone masonry walls: experimental validation using an in-situ testing approach
Core Problem: Stone masonry with mud mortar remains highly vulnerable in Himalayan school buildings and requires low-cost validated retrofit methods.
Key Innovation: Uses in-situ testing to validate welded-wire-mesh strengthening for stone masonry walls, supporting practical seismic resilience in Nepal.
26. GeoHazards, Vol. 7, Pages 72: Communicating the Last Mile of Seismic Risk: Insights from a Case Study
Core Problem: Earthquake risk communication often stops at event parameters and structural collapse while neglecting local site effects and non-structural vulnerability.
Key Innovation: Evaluates an experience-based exhibition that targets actionable public understanding of local seismic risk and non-structural safety.
27. Potential Flood Events in the Lower Missouri River Basin Over Multiple Centuries Identified Using Tree-Ring Based Multi-Model Streamflow Reconstructions
Core Problem: Instrumental records under-sample rare high-flow events in the Missouri River Basin.
Key Innovation: Uses tree-ring-based streamflow reconstructions and multiple models to extend potential flood-event identification across several centuries.
28. Urban Waters, Unseen: The Hydrology of Informal Settlements Must Not Be Ignored
Core Problem: Informal settlements are hydrologically active but are commonly absent from urban flood and water-risk models.
Key Innovation: Argues for community sensing, tailored remote sensing, and settlement-specific hydrological representation to reduce blind spots in urban risk assessment.
29. Advancing green tide monitoring: a spatiotemporal analysis in the Southern Yellow Sea using a novel deep learning and remote sensing fusion approach
Core Problem: Green tides in the Southern Yellow Sea require consistent multi-temporal monitoring despite filamentous morphology and scene-dependent spectral variability.
Key Innovation: Develops UPS-Net-LDU and a 2018-2025 GF-1 WFV dataset to map green-tide distributions with edge-enhanced and attention-guided segmentation.
30. Upstream and downstream channel responses to bridge construction in the Godavari River, India: evidence from multi-temporal geospatial analysis
Core Problem: Bridge construction can reorganize flow, sediment pathways, and local erosion, yet long-term upstream and downstream adjustments are not well quantified.
Key Innovation: Uses multi-temporal Landsat imagery and DEM analysis to document two decades of channel response at three Godavari River bridge sites.
31. Efficient surrogate analysis method for tunnel surrounding rock deformation prediction based on mechanical parameter inversion
Core Problem: Real-time tunnel deformation forecasting is limited by numerical cost, parameter uncertainty, and poor transfer from excavated to unexcavated sections.
Key Innovation: Combines mechanical-parameter inversion with a random-forest surrogate optimized by an improved sine-cosine and Cauchy sparrow search algorithm.
32. Instability Behavior of the Tunnel Face during Shield Undercrossing the Existing Tunnel in Water-Rich Sandy Gravel Strata
Core Problem: Shield tunneling beneath existing tunnels in water-rich sandy gravel can trigger face instability and threaten operating infrastructure.
Key Innovation: Uses coupled CFD-DEM simulations to compare dry and water-rich strata and quantify support pressure requirements during undercrossing.
33. Investigation on failure mechanism of layer surrounding rock of tunnel using a cohesive zone model based on finite discrete element method
Core Problem: Bedding planes in layered rock masses produce complex excavation-induced failure around tunnels.
Key Innovation: Builds a cohesive-zone finite-discrete element framework to reconstruct tension-shear fracture distributions and progressive failure mechanisms during excavation.