TerraMosaic Daily Digest: Mar 15, 2026
Daily Summary
The March 15, 2026 literature is distinguished by an unusually strong convergence around quality-conditioned hazard inference. The landslide papers no longer treat inventories as passive inputs. Instead, they interrogate how canopy cover hides scar geometry, how sample purity governs susceptibility bias, how compact remote-sensing workflows reveal slow-moving slopes before catastrophic acceleration, and how future reservoir, rainfall, and seismic forcing should be coupled within one slope-stability forecast. This is a substantive methodological advance: the problem is shifting from merely selecting better models to building better evidence layers, cleaner supervision, and more physically constrained scenarios.
A parallel development is visible in hydroclimatic and infrastructure hazards. Flood, GLOF, coastal-aquifer, geothermal-fracture, liquefaction, and tunnel studies all emphasize that hazard emerges from interacting states rather than single variables: discharge thresholds, storm-surge recurrence, fracture thermal history, surcharge transfer, and hydraulic connectivity all reshape response. Several papers translate this insight into operational systems, from compact glacial-lake mapping networks to AI weather downscaling for cyclone impacts and engineered tunnel segments that directly reduce overload. The resulting picture is one of geohazard science becoming more transferable, more scenario-aware, and more useful at the point of decision.
Key Trends
Today's strongest papers combine higher-quality observations with coupled forcing analysis, making hazard estimates more robust under real-world environmental change.
- Landslide intelligence is becoming quality-driven: sample purity, forest-cover effects, and long-horizon displacement tracking are now treated as first-order controls on susceptibility and detection reliability.
- Future stability is increasingly evaluated under coupled scenarios: rainfall, reservoir impoundment, seismic loading, storm surge, and groundwater response are being modeled jointly rather than sequentially.
- Compact monitoring systems are gaining operational value: feature tracking, compact glacial-lake segmentation, and AI weather correction-downscaling are designed for repeatable deployment in data-limited settings.
- Infrastructure research is shifting toward engineered hazard reduction: surcharge-buffer tunnel segments, deformation-control schemes, and physics-informed settlement surrogates are focused on design choices, not just diagnosis.
- Multi-hazard comparison is becoming more decision-ready: national-scale risk baselines and asset-scale impact models now place floods, landslides, earthquakes, wind, and cryosphere hazards on more comparable footing.
Selected Papers
This digest features 22 selected papers from 1084 papers analyzed across landslide detection and susceptibility, critical slope forecasting, urban flood modelling, GLOF surveillance, induced seismicity, coastal hydroclimatic change, and tunnel-geotechnical mitigation.
1. Detection of slow-moving landslides using satellite imagery at the Glacier Bay National Park and Preserve
Core Problem: Slow-moving slopes in remote glaciated terrain remain difficult to detect before they transition into faster and more damaging failure.
Key Innovation: Automated Landsat feature tracking generates velocity maps and displacement time series that reveal 21 Glacier Bay landslides, including 14 previously unknown unstable areas.
2. Numerical Simulation and Stability Forecasting of a Critical Slope Considering Future Reservoir Impoundment under Coupled Hydrological and Seismic Scenarios
Core Problem: Reservoir slopes cannot be forecast credibly when rainfall, water-level rise, and seismic forcing are evaluated in separate silos.
Key Innovation: Field-constrained FEM simulations show how impoundment, pore-pressure rise, rainfall, and earthquake loading collapse slope safety in coupled scenarios near Dasu Reservoir.
3. Enhancing landslide susceptibility mapping through a hybrid model utilizing bivariate methods and convolutional neural networks
Core Problem: Deep-learning susceptibility maps often improve predictive skill while sacrificing interpretability in mountainous terrain.
Key Innovation: A hybrid IV-CNN workflow improves both AUC and transparency by combining bivariate statistics with convolutional learning for Himalayan landslide susceptibility mapping.
4. Influence of sample purity on landslide susceptibility mapping: a mutually exclusive stratified purity strategy based on the state-process consistency
Core Problem: Low-purity training samples introduce generalized bias and conservative misclassification in machine-learning susceptibility maps.
Key Innovation: A state-process-consistent purification framework sharply improves sample representativeness and produces more balanced, higher-confidence landslide zonation.
5. Topographic profile and morphology analysis of shallow landslides inside and outside of forests with a semi-automatic mapping approach and bi-temporal airborne laser scanning data
Core Problem: Inventories that ignore forest-covered landslides systematically distort scar geometry and failure characterization.
Key Innovation: Bi-temporal ALS and semi-automatic scar mapping reveal that forested shallow landslides are generally deeper, thicker, and steeper than open-slope failures.
6. Comprehensive multi-hazard risk assessment in data-scarce regions – a study focused on Burundi
Core Problem: Data-scarce countries often lack a common framework for comparing losses across floods, landslides, earthquakes, winds, and torrential rains.
Key Innovation: A nationwide framework for Burundi harmonizes hazard, exposure, and vulnerability into comparable annual-average-loss metrics, including a national shallow-landslide susceptibility layer.
7. Frictional Heterogeneity Governs Slip Partitioning and Seismic Hazard in the 2023 Turkey Earthquake Doublet
Core Problem: Seismic rupture, afterslip, and aftershock partitioning remain difficult to explain without direct constraints on spatial frictional variability.
Key Innovation: Postseismic InSAR inversion within a rate-and-state framework maps where velocity-strengthening versus unstable patches governed slip behavior in the 2023 Turkey doublet.
8. Simulation of flood associated with extreme rainfall events over the urban city, Bengaluru, India
Core Problem: Bengaluru's flood sensitivity under repeated extreme-rainfall events remains poorly constrained despite rapid urban encroachment on natural detention areas.
Key Innovation: SWMM simulations identify rainfall thresholds for widespread inundation and demonstrate how low-lying urbanized zones near historic lakes dominate flood response.
9. Mapping glacial lakes with glacial-lake compact efficient neural network (GCE-Net): perspectives on balancing model size and performance
Core Problem: Frequent glacial-lake mapping is needed for GLOF surveillance, but large segmentation models remain costly for repeated deployment in mountain regions.
Key Innovation: GCE-Net delivers compact, prunable lake segmentation with high accuracy and supports a 2023 glacial-lake inventory across the southeastern Tibetan Plateau.
10. Climate Change Alters Post‐Surge Recovery of Coastal Aquifers
Core Problem: It remains unclear when repeated storm surges allow coastal aquifers to recover and when they force a shift toward persistent salinization.
Key Innovation: Integrated simulations define recovery versus shifted-equilibrium regimes and tie long-term salt accumulation to a compact surge-frequency–intensity metric.
11. Cyclic shear behavior of thermally treated sandstone fractures under constant normal stiffness boundary conditions
Core Problem: Injection-driven slip in geothermal reservoirs cannot be managed well without understanding how fracture behavior evolves with thermal history and shear cycling.
Key Innovation: CNS cyclic-shear tests reveal temperature thresholds and a shift from cohesion- to friction-dominated fracture response under repeated loading.
12. Durability of PA-assisted MICP-reinforced sand in marine environment
Core Problem: MICP-reinforced sand degrades under marine exposure, limiting its use as a durable erosion-control intervention.
Key Innovation: Polycarboxylic-acid assistance improves both initial strength and seawater durability, while secondary treatment restores or exceeds post-exposure strength.
13. Micro-mechanical interpretation on variation of liquefaction resistance of granular materials with different levels of cyclic pre-shearing
Core Problem: Variations in liquefaction resistance after cyclic pre-shearing are well observed but poorly explained by standard fabric indicators.
Key Innovation: DEM analysis shows that tangential-force anisotropy and the sliding potential of sticking contacts, rather than conventional coordination metrics, control the resistance shift.
14. From AI Weather Prediction to Infrastructure Resilience: A Correction-Downscaling Framework for Tropical Cyclone Impacts
Core Problem: Fast AI weather models remain too coarse for asset-scale infrastructure warning during tropical cyclones.
Key Innovation: ACDF converts coarse AI weather forecasts into 500 m wind fields and transmission-line failure probabilities, enabling tower-scale impact forecasting in seconds.
15. Temperature sensitivity and rainfall heat flux drive rapid mass loss of low-latitude glaciers in the Southeastern Qinghai–Tibet Plateau
Core Problem: The joint contribution of rainfall heat flux and warming to rapid mass loss remains poorly resolved for low-latitude glaciers.
Key Innovation: Surface-energy-balance analysis clarifies how rain-on-snow heat input and temperature sensitivity combine to accelerate glacier retreat in the southeastern Tibetan Plateau.
16. Mechanisms of saltwater intrusion and its sensitivity to river discharge in a medium-sized estuary: Insights from the Minjiang Estuary, China
Core Problem: Medium-sized estuaries are highly vulnerable to freshwater stress, yet their salt-intrusion dynamics remain understudied relative to large estuaries.
Key Innovation: Field observations and 3D modelling show that the Minjiang Estuary is exceptionally discharge-sensitive and resolve the branch-scale mechanics of wet- and dry-season intrusion.
17. Study on deformation control technology for bilateral deep foundation pit construction in metro station
Core Problem: Bilateral deep foundation pits beside metro facilities can generate dangerous uplift and lateral deformation during excavation.
Key Innovation: Numerical simulation and field validation show that integrated control measures can substantially reduce tunnel and station deformation in asymmetric urban excavation.
18. Effect of overlying low-permeable soil on underwater shield tunnel response to water level fluctuation
Core Problem: Low-permeability cover above underwater tunnels delays pore-pressure transmission and complicates prediction of lining response to water-level change.
Key Innovation: Experiments and analytical modelling quantify how clay cover amplifies pressure heterogeneity and increases tunnel deformation under fluctuating river levels.
19. An EPS-NC composite segment for mitigating surcharge-induced loads on shield tunnels: Experimental and theoretical analysis
Core Problem: Surcharge loading in reclaimed ground can impose excessive vertical pressure and deformation on shield tunnels.
Key Innovation: EPS-NC composite segments function as a compressible buffer, sharply reducing tunnel-crown load while keeping added settlement within a controlled range.
20. Flow Structures and Their Controls on Downstream Discharge Regulation for a Combined Confluence and Bifurcation Network Node
Core Problem: H-shaped confluence-bifurcation nodes can regulate downstream flood flows, but the controlling flow structures remain poorly understood.
Key Innovation: Hydrodynamic simulations show how upstream discharge ratio reverses connecting-channel flow and creates stabilizing structures that act like natural diversion facilities.
21. Extreme drought–accelerated dissolved carbon metabolism triggers pulsed CO2 outgassing in karst lakes
Core Problem: Extreme drought alters dissolved-carbon cycling in karst lakes, but the timing and mechanism of resulting CO2 pulses are not well constrained.
Key Innovation: Field observations show that drought accelerates organic-carbon respiration and primes post-drought pulses of CO2 outgassing when wetter conditions return.
22. Efficient modeling of Pile-Tunnel Interaction: A PINN-Based framework for vertical settlement prediction
Core Problem: Settlement prediction for existing tunnels under nearby pile loading is often either too slow numerically or too weak physically in purely data-driven models.
Key Innovation: A physics-informed neural network embeds the Timoshenko-Pasternak equation into settlement prediction, improving computational efficiency without discarding mechanics.