TerraMosaic Daily Digest: June 22, 2026
Daily Summary
The June 22 literature is narrower than the preceding issue but scientifically useful: it shifts attention from large landslide inventories to process evidence, impact mechanics, and disaster-specific remote sensing. The strongest geohazard contribution is a debris-flow study that tracks how building fragments are buried and displaced, linking flume observations with an analytical force balance that matters directly for rescue search spaces. A second process strand connects soil moisture memory, sediment connectivity, rock-slope parameter sensitivity, and post-cyclic soft-clay response, showing how hydrological persistence, human modification, damaged rock masses, and sequential seismic loading condition later failure.
The remote-sensing papers are most compelling where they target operational hazard evidence rather than generic mapping. Vision foundation models are adapted to centimetre-scale floodwater mapping; RGB and multispectral data are fused for post-earthquake building-damage segmentation; Sentinel-1 SAR is used to quantify Amazon river-planform dynamics; European Ground Motion Service data reveal peat-like displacement behavior; and GNSS constraints are used to frame segment-scale strain on the central North Anatolian Fault. These papers do not form one method family, but they share a practical goal: converting dense observations into interpretable hazard or exposure variables.
Earthquake and flood risk papers add the consequence layer. Mainshock-aftershock analysis and repair-cost assessment from the 2023 Türkiye earthquake sequence link structural response to cumulative damage and economic recovery. Flood-frequency inference, extreme-rainfall pattern recognition, sand-spit monitoring, rainfall erosivity projections, built-environment combustible-mass mapping, and UAV fire-spread quantification broaden the issue from slope failure to compound hydroclimatic and fire-related risk. The day therefore reads as a methods-and-evidence digest: fewer landmark landslide papers, but a strong set of tools for measuring the precursors, impacts, and uncertainty that make geohazards operationally hard.
Key Trends
Five movements define this issue: state-based process monitoring, event-specific disaster remote sensing, consequence-aware seismic analysis, hydroclimatic geomorphology, and constrained use of foundation models.
- Process measurements are replacing static labels: Soil moisture memory, debris-flow fragment transport, sediment connectivity, and disturbed-rock stability papers all emphasize state, timing, and mechanism rather than one-off hazard classification.
- Disaster remote sensing is becoming event-specific: Floodwater mapping, post-earthquake damage segmentation, SAR river-planform monitoring, EGMS displacement mining, and GNSS strain analysis convert remote observations into event-scale hazard evidence.
- Seismic risk papers are moving from shaking to consequences: The Türkiye mainshock-aftershock and repair-cost studies connect ground-motion sequences to cumulative damage, residual performance, reconstruction burden, and financial exposure.
- Hydroclimatic hazards are treated as coupled geomorphic problems: Sand-spit dynamics, flood-frequency inference, monsoon rainfall recognition, and rainfall erosivity projections link hydrology to sediment, morphology, infrastructure, and future risk.
- Foundation models enter the digest only when tied to a hazard observable: The strongest AI papers are not generic architecture papers; they adapt foundation or transformer models to flood extent, building damage, SAR change, or other measurable disaster states.
Selected Papers
The selected papers emphasize debris-flow impact mechanics, soil-moisture memory, rock-slope stability, floodwater mapping with vision foundation models, post-earthquake building-damage segmentation, SAR river monitoring, GNSS-constrained fault strain, earthquake loss evidence, flood-frequency inference, wildfire exposure data, UAV fire-spread quantification, InSAR ground deformation, rainfall erosivity, and cold-region geotechnics. This issue contains 27 selected papers from 2335 papers analyzed.
1. Buried and displaced: moving characteristics of building fragments in debris flows
Core Problem: Buildings destroyed by debris flows or flow-type landslides can be transported and buried far from their original locations, making post-event rescue search zones difficult to constrain.
Key Innovation: Combines controlled flume experiments, embedded inertial measurements, and an analytical model of drag, earth pressure, and basal friction to explain building-fragment motion in debris-flow deposits.
2. Scale-dependent transition in soil moisture memory and its environmental controls in complex mountain terrain
Core Problem: Soil moisture persistence controls hydrological response and geohazard susceptibility, but its timescale dependence in landslide- and debris-flow-prone terrain remains poorly resolved.
Key Innovation: Uses 20 years of daily soil-moisture data, spectral persistence metrics, detrended fluctuation analysis, and scale-dependent attribution to quantify soil-moisture memory across erosion, landslide, and debris-flow catchments.
3. Parameter measurement sensitivity in disturbed rock masses: A machine learning framework for slope stability prediction
Core Problem: Rock-slope factor-of-safety estimates can be biased when disturbance-induced strength degradation and failure-criterion choice are simplified.
Key Innovation: Builds a finite-element and machine-learning framework over 432 simulations to quantify sensitivity to GSI, disturbance factor, and failure criterion in reinforced disturbed rock slopes.
4. Flood Mapping from RGB imagery using a Vision Foundation Model
Core Problem: Emergency flood mapping from airborne RGB imagery needs models that transfer across events despite limited labels and differences from satellite pretraining data.
Key Innovation: Adapts the Prithvi-EO-2.0 foundation model with a UPerNet decoder for centimetre-scale RGB floodwater segmentation and tests cross-event transfer on BlessemFlood21 and NeuenahrFlood imagery.
5. MS-SwinNet: Early Fusion of RGB and Multispectral Imagery for Postearthquake Building Damage Segmentation
Core Problem: Post-earthquake urban scenes contain fragmented damage, class imbalance, and confusing materials that make damaged buildings difficult to segment from single-modality imagery.
Key Innovation: Introduces a shifted-window transformer network that fuses RGB and eight-band multispectral data at the input stage for damage segmentation after the 2023 Kahramanmaraş earthquake sequence.
6. Monitoring sand spit variability using PlanetScope satellite imagery in a high-energy microtidal urbanized estuary: hydrodynamic drivers and flood risk management implications for central Chile
Core Problem: Urbanized estuary sandbars can close or breach under flood and wave forcing, changing flood-risk pathways over event-scale timescales.
Key Innovation: Uses multitemporal PlanetScope classification and machine-learning attribution to relate Maipo estuary sand-spit morphology to wave power and river discharge.
7. Anthropogenic control on sediment connectivity for soil resource management in sloping vineyards (Mercurey, France)
Core Problem: Sloping vineyards can generate soil loss and mudflows, but flood-management structures modify sediment routing in ways that are not captured by standard erosion models.
Key Innovation: Combines InVEST SDR, RUSLE-based erosion estimates, connectivity indices, and sediment-trap efficiency corrections to quantify how anthropogenic structures interrupt sediment delivery.
8. Seismic performance of reinforced concrete structural systems under mainshock-aftershock sequences: insights from the 2023 Türkiye-Syria earthquakes
Core Problem: Conventional seismic design often evaluates single-event loading, leaving cumulative damage from earthquake sequences underrepresented.
Key Innovation: Applies recorded 2023 Türkiye-Syria mainshock-aftershock motions to prototype RC buildings to quantify cumulative damage, stiffness loss, residual drift, and collapse risk.
9. Assessment of repair/reconstruction costs to existing reinforced concrete and masonry buildings following the 2023 Kahramanmaraş earthquakes
Core Problem: Seismic resilience assessment requires cost evidence from real damaged building stock, not only structural fragility curves.
Key Innovation: Analyzes TCIP damage and compensation data for more than 450,000 units after the 2023 Kahramanmaraş earthquakes to evaluate repair and reconstruction cost patterns.
10. Flood frequency analysis combining observed streamflows with simulations from a continental hydrological model
Core Problem: Flood-frequency estimates at sites with short streamflow records remain highly uncertain, especially where long observational baselines are unavailable.
Key Innovation: Introduces CoFFI, a censored bivariate Bayesian inference method that combines short observed annual maxima with long continental-model simulations through a copula formulation.
11. Assessment of River Planform Dynamics in the Amazon Basin Using Sentinel-1 SAR Data (2017-2025)
Core Problem: River-planform change affects infrastructure, food security, and ecosystem stability, but cloud-independent basin-scale monitoring remains difficult in the Amazon.
Key Innovation: Uses quarterly Sentinel-1 C-band SAR composites, water-mask generation, and river-metric extraction to map Amazon planform dynamics from 2017 to 2025.
12. Segment-Scale Strain Accumulation and Seismic Potential of the Central North Anatolian Fault Zone with GNSS Constraints
Core Problem: Seismic potential along major strike-slip faults depends on how strain accumulation varies between fault segments.
Key Innovation: Uses GNSS constraints to estimate segment-scale strain accumulation along the central North Anatolian Fault Zone and evaluate its seismic potential.
13. How do geoenvironmental fire disasters impact land surface thermal budget and threaten biophysical equilibrium?
Core Problem: Wildfires alter land-surface thermal budgets and biophysical equilibrium, but post-fire impact assessment often relies on isolated spectral indices.
Key Innovation: Builds a geospatial multi-model workflow using land-surface temperature, burn indices, greenness-drought disturbance models, and threshold segmentation to assess fire-triggered environmental damage.
14. COMBUST: Gridded combustible mass estimates of the built environment in the conterminous United States (1975-2020)
Core Problem: Wildfire and conflict-related fire-risk assessment needs spatially explicit estimates of combustible mass in the built environment, not only vegetation fuel maps.
Key Innovation: Provides gridded built-environment combustible-mass estimates for the conterminous United States from 1975 to 2020 to support risk and damage assessment.
15. Quantifying grassland fire spread via UAV videogrammetry: boundary-precise segmentation and EDT-based vector field estimation
Core Problem: Tactical wildfire management needs fine-spatiotemporal estimates of fire rate of spread, but segmentation boundary errors can dominate the velocity signal.
Key Innovation: Combines boundary-precise UAV video segmentation with Euclidean-distance-transform vector-field estimation to derive grassland fire spread rates at operational frame rates.
16. Pattern Recognition of West African Monsoon Extreme Rainfall Events Using Convolutional Neural Networks
Core Problem: Extreme rainfall and dry events in West Africa are difficult to classify because monsoon dynamics involve nonlinear moisture-flux and circulation interactions.
Key Innovation: Trains a CNN on vertically integrated moisture-flux convergence and uses interpretability diagnostics to identify physically coherent wet and dry event patterns.
17. Detecting peat-like behaviour areas using European Ground Motion Service data
Core Problem: Peatland-like displacement behavior is difficult to detect with optical mapping alone, especially where land-cover classifications miss subsiding or seasonally deforming ground.
Key Innovation: Uses European Ground Motion Service InSAR time series with PCA and k-means clustering to identify peat-like vertical displacement dynamics across Great Britain.
18. Unravelling the interplay between event frequency and intensity in driving rainfall erosivity future changes
Core Problem: Future soil-erosion risk depends on both rainfall intensity and erosive-event frequency, but convection-permitting climate simulations are still underused in erosivity studies.
Key Innovation: Bias-corrects hourly convection-permitting climate simulations for Sicily and separates the roles of event frequency and intensity in projected rainfall erosivity change.
19. Tidal sensitivity of tremors in a mixed fast and slow earthquake system in northeastern Japan
Core Problem: Tidal modulation of tremor provides a sensitive probe of fault stress response, but behavior in mixed fast- and slow-earthquake systems remains poorly constrained.
Key Innovation: Analyzes 2016-2024 tremor catalogs in northeastern Japan to assess spatial changes in tidal sensitivity along a subduction-zone system.
20. Post-cyclic simple shear responses of soft clay under equivalent sequential seismic loading with various degrees of reconsolidation
Core Problem: Whole-life seismic design of geotechnical systems requires understanding how soft clay recovers or weakens after sequential earthquake loading.
Key Innovation: Performs cyclic and post-cyclic simple-shear tests under equivalent mainshock-aftershock loading and varied reconsolidation to quantify post-earthquake shear response.
21. Prediction of Seismic Earth Pressure on Basement Walls of Buildings with Rigid Base Conditions
Core Problem: Basement-wall seismic design is sensitive to soil-structure interaction and building inertia, which are often simplified in earth-pressure estimates.
Key Innovation: Uses direct numerical simulations to derive an analytical solution for seismic earth pressure on basement walls under rigid and compliant base conditions.
22. Towards the quantification of seismic impedance function using in situ soil-structure interaction measurements
Core Problem: Impedance functions are central to soil-structure interaction analysis, but field-based quantification remains difficult because post-processing forced-vibration measurements is nontrivial.
Key Innovation: Demonstrates a reduced-scale proof of concept for estimating seismic impedance functions from in situ soil-structure interaction measurements.
23. Comparative Assessment of Lead Rubber and Friction Pendulum Seismic Isolation Systems Under Varying Seismic Hazard and Site Conditions
Core Problem: Isolation-system performance varies with seismic hazard level and soil class, but code-based comparisons are often limited to one site condition.
Key Innovation: Compares lead-rubber and friction-pendulum isolation systems across Turkish seismic hazard levels and site classes using displacement demand, base shear, and code checks.
24. Dynamic Mechanical Properties and Constitutive Modeling of Saline Frozen Soil
Core Problem: Frozen saline soils in cold-region infrastructure experience salt-dependent weakening and impact loading that are not captured by generic frozen-soil models.
Key Innovation: Uses dynamic impact tests, energy-dissipation analysis, and a modified Johnson-Cook model to characterize saline frozen clay under varied salt content and loading.
25. Data Driven Modelling of Groundwater and Seawater Intrusion Monitoring, Prediction, and Management in Arid Regions
Core Problem: Arid coastal aquifers face seawater intrusion and overextraction under sparse and noisy monitoring, limiting early warning and management.
Key Innovation: Reviews machine learning, deep learning, and semi-supervised approaches for groundwater and seawater-intrusion monitoring and prediction in data-scarce regions.
26. Vegetation restoration induces the co-evolution of red soil structure and organic carbon on steep terraced slopes
Core Problem: Engineered steep slopes require restoration strategies that stabilize soil aggregates while improving carbon retention, but their coupled evolution is poorly understood.
Key Innovation: Tracks aggregate structure and organic-carbon fractions across restoration patterns and ages to show how vegetation succession improves red-soil stability on steep terraces.
27. CSD-Net: Content-Style Decoupling with Exploratory MLLM-Guided Refinement for Robust Change Detection
Core Problem: Operational change detection for hazards is degraded by radiometric pseudo-changes caused by seasonality, illumination, and atmosphere.
Key Innovation: Introduces content-style decoupling and exploratory MLLM-guided refinement to separate geometric change from environmental style in bi-temporal remote-sensing imagery.