TerraMosaic Daily Digest: September 4, 2026
Daily Summary
Measured deformation substantially changes the hazard picture in this day's literature. Along the Volta Delta, incorporating spatially resolved PS-InSAR subsidence raises the share of coastline rated high to very high vulnerability from 28% to 75%, demonstrating that regional vertical-motion estimates can conceal local exposure. A complementary Houston study turns subsidence history into operational groundwater limits by mapping pre-consolidation head, safe pumping buffer and safe pumping volume; the contrast between an approximately 5 m buffer near NASA-Clear Lake and 24 m near Pasadena shows how sharply pumping headroom can vary within one metropolitan aquifer system. On the Brazilian coast, 93 mapped overwash events connect storm forcing and foredune configuration to a post-2005 phase of accelerated barrier retreat. Together these studies move deformation from a background covariate to a spatially explicit control on risk and intervention.
Flood and infrastructure studies emphasize decision-relevant prediction under nonstationarity. At Camp Creek, five analytical approaches agree that forestry affects peak flow but differ in how risk can be interpreted; design-life level analysis offers a coherent probability measure while the equivalent-clearcut-area covariate remains uncertain. Peak-oriented deep learning similarly broadens evaluation beyond mean error by combining event detection, epistemic and input uncertainty, SHAP attribution and multi-criteria ranking. Differentiable hydrology extends this direction toward cross-basin transfer, while an Australian study directly compares recovery after slow- and rapid-onset floods. In Lisbon, aggregate interaction can make isolated-building seismic assessments either conservative or unsafe. Multiobjective bridge modeling on liquefiable ground further links soil improvement to fragility, recovery, cost and carbon rather than a single performance target.
Across environmental observation and geomaterials, the strongest methodological gains come from representing structure before fitting response. Deep-learning preprocessing of AMSR-E/2 observations reduces soil-moisture reanalysis bias from 28.68% to 3.28%, with spatially varying observation uncertainty improving temporal correlation at 66% of validation stations in a dedicated experiment. High-resolution flow imaging rejects linear resistance superposition in shallow runoff, while paired retention-freezing data support a common fractal description of pore control. Constitutive, tomography and pilot-scale treatment studies resolve how fabric, bonding, bedding, damage and preferential flow govern clay response, shale fracture connectivity, caprock sealing and EICP uniformity. Multiscale sensing and computation, spanning InSAR, satellite-UAV updates, LiDAR benchmarks, lightning fusion and sparse radar reconstruction, expand observational capability, but their transfer to landslide detection or forecasting remains to be demonstrated directly.
Key Trends
The common methodological shift is from static, spatially averaged indicators toward state-aware models that preserve heterogeneity, uncertainty and process coupling.
- Local deformation is becoming a first-class risk variable: The Volta Delta and Houston studies quantify how spatially resolved subsidence changes vulnerability classes and safe pumping thresholds, while the Santa Catarina analysis connects repeated overwash to decadal barrier retreat. Risk estimates become materially different when deformation is measured locally rather than represented by a regional constant.
- Hazard prediction is moving toward uncertainty-aware, decision-specific objectives: Peak-flow learning weights extremes and ranks models across accuracy, event reliability, uncertainty and interpretability. Design-life analysis reframes forestry effects under nonstationarity, and differentiable hydrology seeks transferable process structure for data-scarce basins. The emerging target is not a single fit statistic but a forecast linked to the decision it must support.
- Material structure and interaction replace additive simplifications: Overland-flow experiments show that grain and form resistance interact nonlinearly. Fractal pore geometry links retention and freezing, while clay, shale, caprock and glauconite studies track evolving fabric, bonding, fracture connectivity and stress-path dependence. Similar coupling appears in wave solvers and water-tuned periodic barriers, where boundary interactions determine system response.
- Multiscale sensing is being organized around provenance and physical consistency: PS-InSAR supplies local deformation to coastal indices; satellite and UAV evidence are fused in an incrementally updateable terrain map; and ground LiDAR, lightning networks and incomplete-aperture radar motivate benchmarked or constrained reconstruction. The useful advance is not resolution alone, but retaining where observations came from and what physical structure they preserve.
Selected Papers
The collection is led by direct studies of coastal subsidence, overwash, groundwater-induced compaction and liquefaction-sensitive infrastructure, followed by flood prediction, recovery and loess-slope stabilization. Lower-ranked studies extend the methodological range through hydrological transfer, constitutive mechanics, multiscale remote sensing and constrained reconstruction. Across these fields, the recurring question is whether local variability survives aggregation, from deformation cells and aquifer thresholds to soil fabric and multimodal observations.
1. Sustainability-Driven Seismic Resilience Assessment and Multiobjective Optimization of DSM-Retrofitted RC Bridges on Liquefiable Ground
Core Problem: Deep soil mixing for bridges on liquefiable ground must balance seismic resilience against cost and embodied carbon.
Key Innovation: Couples a nonlinear three-dimensional soil-pile-bridge model to fragility, recovery, cost and life-cycle carbon objectives across alternative DSM configurations.
2. Coastal Vulnerability Index (CVI) Assessment of a Data-Sparse Delta: Quantifying the Contribution of InSAR-Derived Land Subsidence in the Volta Delta, Ghana
Core Problem: Coastal vulnerability indices for data-sparse deltas can severely understate risk when they omit spatially resolved land subsidence.
Key Innovation: Integrates validated PS-InSAR subsidence across 72 coastal cells; high-to-very-high vulnerability rises from 28% without subsidence to 75% with local deformation, with every cell increasing significantly.
3. Overwash processes: spatio-temporal scales and dynamics of the coastal system in southern Santa Catarina, Brazil
Core Problem: Event-scale overwash drivers must be linked to multi-decadal shoreline behavior to explain barrier-system retreat.
Key Innovation: Integrates remote sensing, aerial photographs and ocean-climate records to identify 93 overwash events and relate their recurrence to an intensified transgressive phase after 2005.
4. Mapping new pre-consolidation head, safe pumping buffer, and safe pumping volume for groundwater management in post-subsidence cities: a case study from Houston, Texas
Core Problem: Post-subsidence cities need spatial pumping limits that prevent hydraulic heads from crossing thresholds that reactivate permanent compaction.
Key Innovation: Maps pre-consolidation head, safe pumping buffer and safe pumping volume across Houston, distinguishing a roughly 5 m buffer near NASA-Clear Lake from 24 m near Pasadena.
5. Camp Creek Revisited, One More Time: Evaluating Multiple Approaches to Quantify Peak Flow Response to Forestry for a Medium-Size Snow-Dominated Catchment
Core Problem: Peak-flow effects of progressive forest disturbance are difficult to quantify when pre-harvest relations and return periods are nonstationary.
Key Innovation: Compares five inferential frameworks and identifies design-life level analysis as the most meaningful probabilistic formulation, while exposing equivalent clearcut area as a remaining weak link.
6. Peak-oriented one-day-ahead streamflow forecasting using hybrid deep learning: uncertainty quantification, SHAP-based interpretability, and MCDA-based model selection
Core Problem: Daily streamflow models optimized for average error can miss peak magnitude, event detection and operational uncertainty.
Key Innovation: Combines peak-weighted fine-tuning, event diagnostics, Monte Carlo dropout, meteorological perturbations, SHAP and multi-criteria ranking; U-CNN-BiGRU leads with an integrated score of 0.69.
7. Contrasting Community Experiences and Recovery Challenges in Slow- and Rapid-Onset Flooding: Insights from the 2022-2023 Australian Floods
Core Problem: Flood onset rate may shape community experience and recovery needs in ways that are obscured when events are treated as a single hazard class.
Key Innovation: The verified title frames a direct comparison of slow- and rapid-onset Australian floods; the accessible publisher record does not expose methods or results, so no finer claim is made.
8. Surrogate regression models for seismic performance assessment of enclosed structural units within masonry aggregates
Core Problem: Treating masonry units as isolated buildings can misrepresent the seismic capacity of enclosed units in historic urban aggregates.
Key Innovation: Derives cloud-based surrogate relations from aggregate archetypes and incremental ground-acceleration analyses, enabling rapid capacity correction without modeling the full building cluster.
9. A differentiable Xin’anjiang model incorporating transfer learning for cross-basin flood prediction under data scarcity
Core Problem: Flood models trained basin by basin cannot readily transfer process knowledge where observations are limited.
Key Innovation: The verified title couples a differentiable Xin'anjiang model with transfer learning for cross-basin prediction; the publisher record does not yet expose the journal abstract or evaluation results.
10. Dual-layer stabilization system for loess slopes with performance evaluation and parameter optimization
Core Problem: Loess slopes require protection that jointly addresses near-surface degradation and deeper stability while remaining tunable to site conditions.
Key Innovation: The verified record introduces a dual-layer stabilization system with performance evaluation and parameter optimization; construction details and quantitative gains are not exposed in the available abstract record.
11. A Study on the Applicability of the Linear Superposition Principle to Overland Flow Resistance Based on HR-PIV Technology
Core Problem: Linear addition of grain and form resistance can underestimate resistance and turbulence in shallow overland flow.
Key Innovation: High-resolution PIV demonstrates nonlinear interaction between resistance components and supports a dimensionless product model with R² = 0.86.
12. The Role of Assimilating Deep Learning-Enhanced AMSR-E/2 Observations in European Soil Moisture Reanalysis
Core Problem: Sensor discontinuities, retrieval noise and spatially varying uncertainty can propagate artifacts into continental soil-moisture reanalyses.
Key Innovation: Assimilates a seamless DL-enhanced 2003-2023 AMSR-E/2 record with bias correction and spatial uncertainty, reducing PBIAS from 28.68% to 3.28% and RMSE from 0.100 to 0.082 against in-situ data.
13. A factorial spatial-temporal null-model framework for satellite-ground lightning fusion across Chinese marginal seas
Core Problem: Satellite and ground lightning networks sample storms differently across space and time, complicating fused climatologies over marginal seas.
Key Innovation: The verified title specifies a factorial spatial-temporal null-model framework for testing fusion behavior; numerical performance is not exposed in the available record.
14. Surface-wave attenuation by a composite periodic vibration isolation barrier tuned by mutually matched water levels
Core Problem: Periodic barriers must be tuned to attenuate surface waves over useful frequency bands under variable ground-system conditions.
Key Innovation: The verified title identifies a composite periodic barrier tuned through mutually matched water levels; the accessible record does not expose its attenuation bands or validation results.
15. A Fractal Link Between Soil Water Retention and Freezing
Core Problem: Water retention and soil freezing are usually modeled separately despite their shared dependence on pore structure.
Key Innovation: Derives a fractal soil-freezing characteristic curve and shows comparable pore-structure dimensions across 129 freezing data sets and paired retention-freezing measurements for 14 materials.
16. Thermodynamics-Based Two-Surface Model for Natural Clays Considering Bonding Degradation and Fabric Evolution
Core Problem: Natural clay models must represent plastic deformation inside the conventional yield surface as bonding degrades and anisotropic fabric evolves.
Key Innovation: Combines inner and outer surfaces with a fabric tensor and bond variables in a nine-parameter model validated against monotonic, cyclic and post-cyclic responses of three natural clays.
17. Effect of Glauconite Content on the Compression and Shear Behavior of Sand
Core Problem: The amount of crushable glauconite needed to shift sand from dilative to contractive behavior has not been isolated systematically.
Key Innovation: Controlled mixtures and PIV identify a wet-condition transition between roughly 25% and 50% glauconite and resolve how stress, moisture and interface roughness alter the mechanism.
18. Pilot-Scale Soybean Crude Urease-Based EICP for Soil Improvement in Sand: Distribution and Performance Using Commercial-Grade Reagents
Core Problem: Laboratory EICP performance does not establish whether commercial-grade reagents can produce useful and spatially uniform cementation at pilot scale.
Key Innovation: A 1 m³ injection trial maps 64 locations in three dimensions, reaching 497 kPa maximum compressive strength and showing that preferential flow and injection layout govern treatment uniformity.
19. A two-way multi-fidelity coupling approach combining Boussinesq and Navier-Stokes models
Core Problem: Large-domain wave propagation and localized wave-structure interaction require different model fidelities that must exchange information consistently.
Key Innovation: Uses overlapping subdomains and relaxation-based source terms for non-intrusive bidirectional coupling between Boussinesq and Navier-Stokes solvers.
20. Satellite-UAV Collaborative Off-Road Traversability Mapping and Incremental Updating for Unmanned Ground Vehicles
Core Problem: Regional remote-sensing priors may miss local terrain changes needed for safe off-road route planning.
Key Innovation: Organizes multiscale evidence in a provenance-preserving H3 map and applies confidence-based updates from UAV observations, achieving 96.96% cell-level F1 in the reported test.
21. Quantitative Investigation on the Fracture Evolution of Shale at High Temperatures up to 500℃ via In-situ X-Ray Computed Tomography
Core Problem: The transition from natural-fracture opening to thermally generated connected networks in heated anisotropic shale is poorly quantified.
Key Innovation: Tracks three-dimensional fracture evolution to 500°C and links the mechanism shift from deformation to organic-matter pyrolysis with bedding-dependent network connectivity.
22. Hydromechanical Behaviour of Intact and Recompacted Crushed Caprock for Underground CO₂ Storage
Core Problem: Damaged or crushed caprock near wells and faults may recover density without recovering intact stiffness, permeability or sealing behavior.
Key Innovation: Directly compares intact, pre-fissured and recompacted Opalinus Clay using mechanical, permeability, breakthrough-pressure and X-ray tomography measurements.
23. Mapping forest structure: A 3D deep learning benchmark for ground-based LiDAR
Core Problem: Ground-based LiDAR models for forest-structure mapping lack a common benchmark across architectures and acquisition settings.
Key Innovation: The associated open repository provides model outputs, configurations and evaluation results for point-cloud semantic segmentation using terrestrial and mobile LiDAR.
24. Anthropogenic dominance of water storage variability in the Yellow River Basin: A machine learning synthesis of multi-source data (1981-2031)
Core Problem: Long-term water-storage variability in the Yellow River Basin reflects intertwined climate and human controls that are difficult to separate.
Key Innovation: The verified title identifies a multi-source machine-learning synthesis spanning 1981-2031; the accessible record does not provide enough detail to assert attribution magnitudes.
25. Numerical simulation of oscillatory flow around a circular cylinder at high Reynolds numbers
Core Problem: Resolving oscillatory flow around cylindrical structures at high Reynolds number remains computationally demanding.
Key Innovation: The verified title identifies a high-Reynolds-number numerical study; the accessible publisher record does not expose the numerical scheme or reported results.
26. GPU-Based Solar Irradiance Estimation over Digital Surface Models Using Structurally Lossless Viewshed Compression
Core Problem: Explicit viewsheds make reflective irradiance modeling over large digital surface models memory-bound.
Key Innovation: A structurally lossless visibility encoding reaches about 3.3× compression and keeps a 29.6 GB viewshed resident on a 24 GB GPU, reducing the reported runtime from 6.7 h to 0.5 h.
27. Non-Convex Joint Sparse and Low-Rank Optimization for Enhanced ISAR Imaging from Incomplete Data
Core Problem: Incomplete ISAR observations produce defocusing, while convex approximations can bias sparsity and underuse low-rank scene structure.
Key Innovation: Jointly imposes non-convex sparsity and low-rank constraints and solves the reconstruction with reweighting, truncated SVD and ADMM.
28. CGHU-CD: A contrast-guided hierarchical unsupervised method for detecting Martian anthropogenic changes
Core Problem: Detecting sparse anthropogenic change on Mars without labels requires separating true change from cross-temporal appearance differences.
Key Innovation: The verified title identifies contrast guidance and hierarchical unsupervised inference; the available record does not support claims about architecture or accuracy.
29. Climate change impacts on alpine springs in Austria: shifts in discharge seasonality and water availability
Core Problem: Climate-driven changes in snowmelt and low-flow timing can destabilize spring-water availability across hydrogeologically diverse Alpine catchments.
Key Innovation: A corresponding 2026 conference study reports rainfall-runoff and snow modeling for 76 Austrian springs across three climate scenarios, resolving group-specific shifts in spring and summer discharge.
30. Performance evaluation of pond ash-GGBS stabilized pavement subbase using dynamic cone penetration index
Core Problem: Pond-ash and GGBS subbase systems require field-oriented performance evaluation beyond conventional laboratory strength indices.
Key Innovation: The verified title identifies dynamic cone penetration as the evaluation route; the supplied RSS abstract belongs to a different CBR-prediction study, so no mix design or performance result is inferred.