For deformation monitoring of creeping landslides, MT-InSAR is often limited by phase unwrapping, atmospheric correction, and spatiotemporal filtering, leading to complex workflows and elevated costs. This study proposes a wrapped-interferogram-driven framework for creeping landslide hazard identification and temporal evolution reconstruction. For single-scene recognition, we build the Jinsha Dataset and introduce an Adaptive Frequency-Domain Decoupling and Fusion Block (AFDFB) into the shallow encoder of a segmentation network to mitigate frequency mixing. By coupling convolution with State Space Model (SSM), AFDFB enhances the characterization of high-frequency fringe boundaries and low-frequency structural trends. For temporal modeling, we migrate time-series concepts from the phase domain to the pixel-wise probability domain and integrate coherence weighting, cyclic-consistency constraints, and temporal accumulation to produce an unwrapping-free quasi-time-series risk representation. The proposed network achieves 80.79% IoU and 89.37% F1-score on the test set. In both single-landslide assessment and regional inventory mapping, probability-domain fusion suppresses isolated anomalous responses and yields more stable risk maps, with spatial delineation and temporal variation trends consistent with MT-InSAR.
How to cite: Zhang, R. and Zhu, W.: Unwrapping Is Not All You Need for Capturing the Temporal Evolution of Creeping Landslides, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3298, https://doi.org/10.5194/egusphere-egu26-3298, 2026.
Accurate landslide segmentation from multitemporal remote sensing remains challenging due to strong class imbalance, heterogeneous background disturbances, and ambiguous boundaries induced by shadows, vegetation dynamics, and sensor noise. We propose Change-driven Hysteresis and Inference-ensemble Pseudo-labeling for Landslide Segmentation (CHIPS), a semi-supervised framework that leverages change cues to scale high-quality supervision while controlling error propagation in pseudo-label learning. CHIPS uses LandTrendr-derived spectral change descriptors as the primary representation and couples them with a hysteresis-based selection strategy that assigns pseudo-labels via dual thresholds, enabling confident positives and confident negatives while deferring uncertain pixels to an ignored set. This design explicitly balances precision–recall trade-offs and mitigates confirmation bias under severe foreground sparsity. To further stabilize learning, we integrate an inference-ensemble mechanism that aggregates multiple stochastic predictions (e.g., perturbation- or dropout-based) to estimate pixel-wise confidence and improve pseudo-label reliability. A teacher–student training scheme with exponential moving average supervision combines supervised segmentation loss with pseudo-label and consistency objectives under a scheduled ramp-up. Experiments on a large-scale landslide dataset constructed from change-map patches demonstrate that CHIPS consistently improves intersection-over-union and boundary delineation over fully supervised baselines and common semi-supervised alternatives, particularly in challenging terrain and low-label regimes. The proposed framework offers a practical and scalable solution for regional landslide mapping using change-driven priors and robust pseudo-labeling.
How to cite: Wei, R. and Li, Y.: Change-driven Hysteresis and Inference-ensemble Pseudo-labeling for Landslide Segmentation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6265, https://doi.org/10.5194/egusphere-egu26-6265, 2026.
Critical transport routes in tectonically active regions, such as the Arifiye-Bozüyük corridor in Western Türkiye, are under serious threat due to slope instabilities that are worsened by rugged terrain and complex lithological units. Connecting the Sakarya basin to the Central Anatolian plateau, this region hosts a dense multi-infrastructure network comprising the D-650 Highway (handling over 30,000 vehicles daily), a vital high-speed and conventional railway line, and expanding settlement clusters. Due to the corridor’s proximity to the North Anatolian Fault Zone (NAFZ) and its significance as a primary logistical route, slope failures pose both a local safety risk and a broader threat to national supply chains. A preliminary framework is presented to assess landslide hazard and quantify the exposure of these critical assets. A GIS-based multi-criteria decision analysis was implemented to evaluate landslide hazard by integrating nine causative factors: slope, aspect, curvature, lithology, drainage density, fault density, topographic wetness index (TWI), and distance to road. These parameters were standardized and weighted using the Analytic Hierarchy Process (AHP), with an emphasis on morphological factors and lithological resistance, based on regional expert insights, to capture the specific landslide mechanism in the Arifiye-Bozüyük corridor. The model produced initial hazard zones categorized from low to very high susceptibility. Moving beyond traditional pixel-based susceptibility mapping, the hazard rasters were overlaid with vector-based linear transport networks and building footprints extracted from the Microsoft Planetary Computer open data archive. The object-based approach enabled detailed intersection analysis, distinguishing between overall areal risks and specific infrastructure exposures. The analysis facilitated a comprehensive exposure assessment, pinpointing the spatial distribution of at-risk highway sections, railway segments, and residential structures. An approach that combines AHP-based hazard models with global open-source object data provides a scalable and cost-effective method for initial risk screening. The findings serve as a foundational layer for decision-makers to prioritize detailed field verification, implement early warning systems, and design site-specific geotechnical mitigation measures for the most vulnerable segments.
How to cite: Zengin, E., Ündül, Ö., Doğu, M. M., and Korkut, M.: Preliminary Landslide Hazard and Multi-Infrastructure Exposure Assessment along the Arifiye-Bozüyük Corridor in Western Türkiye, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11644, https://doi.org/10.5194/egusphere-egu26-11644, 2026.
Situating at the tectonic plate junction, the geological conditions in Taiwan is fragile and unstable. The heavy rainfall of typhoons and frequent earthquakes often caused landslide disasters, resulting in significant loss of lives and properties.
Objective of this research aims to construct deep-seated landslide susceptibility model based on discriminant analysis and innovated slope unit. The study area is the Wushe reservoir basin in Central Taiwan with records of frequent deep-seated landslides.
This research uses the slope unit derived using watershed combined with slope aspects as the analysis unit, which has been edited referring to high resolution DEM, aerial photos, and satellite images. The hypothesis testing and Pearson correlation coefficient analysis were conducted on geological, geographical, and hydrological factors derive from DEM. A total of six significant factors were selected to construct the landslide susceptibility model by discrimination analysis. The six significant factors include standard deviation of elevation, river density, average slope angle, geological category, hypsometric integral, and geological structure density. It was found that slope angle and standard deviation of elevation are more important among these factors.
Stratified sampling based on K-mean analysis is carried out based on the standard deviation of elevation and average slope angle, and the discrimination function is used to construct deep-seated landslide susceptibility model in the research area. The ROC curve is used to evaluate the results, and the estimated results compared relatively well with the mapped scarps of the deep-seated landslide.
How to cite: Lin, M.-L. and Huang, P.-Y.: Construction of the deep-seated landslide susceptibility model using watershed-aspect slope unit, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16289, https://doi.org/10.5194/egusphere-egu26-16289, 2026.
Spaceborne Differential SAR Interferometry (DInSAR) is a widely used remote sensing technique to measure Earth surface displacements with high accuracy. The diffusion of DInSAR has been possible thanks to the long-term availability of satellite constellations that regularly acquire SAR data over the Earth at different carrier frequencies and spatial resolutions. Recently, the Copernicus Sentinel-1 mission, which marked its 10th anniversary last year, has further provided the remote sensing community with an unprecedented flood of systematically acquired, format standardized, and open-access data takes. This enabled a shift towards the implementation of monitoring services operating at local and global scale.
In this work, we present an overview of the state-of-the-art of operational DInSAR services aimed at detecting ground displacements induced by various natural (e.g., earthquakes, volcanoes, landslides) and anthropogenic (e.g., gas storage, geothermal exploitation) phenomena at different scales. We also highlight initiatives, primarily the European Plate Observing System (EPOS), that allow the sharing of DInSAR measurements with the wider scientific community, ensuring data reproducibility and knowledge exchange.
Furthermore, we show how the DInSAR measurements can be integrated into civil protection frameworks for hazard evaluation and risk management and mitigation.
Finally, we explore the near-future DInSAR landscape where the availability of the new NISAR, ROSE-L (L-Band) and IRIDE (X-Band) constellations will significantly enhance the ground displacement detection capabilities, further improving our understanding and management of the phenomena under study.
This work has been partly funded by the Italian DPC, in the frame of IREA-DPC (2025–2027) agreement, the HE EPOS-ON (GA 101131592), and the EU-NextGeneratonEU ICSC - CN-HPC - PNRR M4C2 Investimento 1.4 - CN00000013 project.
How to cite: Casu, F.: Monitoring Earth surface displacements through spaceborne radar interferometry techniques, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14507, https://doi.org/10.5194/egusphere-egu26-14507, 2026.
Landslides are significant natural hazards frequently triggered by heavy rainfall and earthquakes, representing the most damaging secondary coseismic environmental effects. In geologically active regions like the Northern Apennines (Italy), high seismic hazard often couples with frequent large-scale slope failures. Global evidence suggests a complex interplay between triggers: earthquakes following intense rainfall tend to induce more landslides, while seismically impacted areas often show elevated landslide rates in subsequent years. Analyzing these tectonic-meteorological interactions is crucial for accurate hazard prediction.
The central goal of this research is to resolve the intricate interactions among tectonic, meteorological, and surface processes by evaluating the role of seismicity and rainfall (whether concurrent or not) in the evolution of slope failures. This presentation details the conceptual framework and preliminary implementation of a newly initiated project aimed at monitoring these dynamics in real-time. The investigation focuses on the fundamental mechanisms of landslide induction, considering pre- and post-seismic meteorological states to identify crucial triggering parameters.
The study utilizes a dedicated, multi-technique monitoring network at the Roncovetro landslide, a relatively young complex-earthflow, with a mean discharge rate of ∼ 0.16 × 105 m3/yr, that serves as a natural laboratory for landslide characterization in the Apennines. To discriminate between induction mechanisms, we integrate:
- Remote Sensing Tools: Repetitive Unmanned Aerial System (UAS) surveys are employed to conduct high-resolution terrain analysis and quantify volumetric changes. Comparison of digital topography (including historical 1973 data vs. 2014–2025 datasets) allows for the assessment of long-term discharge rates and morphological evolution.
- Ground-Based Monitoring: An already existing local network of Global Navigation Satellite System (GNSS) stations and Ultra Wide Band (UWB) sensors provides high-frequency displacement data, enabling the correlation of movement with specific triggers. New GNSS stations will be installed in different sectors of the landslides in order to extend the real time analysis of the slope movements.
- Meteorological Data: Continuous hydro-meteorological parameters are gathered from a nearby weather station managed by the Regione Emilia Romagna, providing the high-resolution rainfall data necessary to establish triggering thresholds.
- Novel Geophysical Sensing: High-resolution seismic data will be acquired through Distributed Acoustic Sensing (DAS), leveraging fiber optic cables to create a dense linear array of seismic sensors at a 1-meter spatial scale.
- Field Analysis: Conventional geomorphological mapping and field-based geological surveys validate the remote sensing products and ground-truth the internal boundaries of the landslide body.
The availability of this integrated observational network will allow for the spatial and temporal discrimination of landslide sectors triggered by meteorological events versus those sensitive to seismic shaking. Future analysis of the Roncovetro site—an area already characterized by historical data and impacted by both significant earthquakes (e.g., the 1996 Mw 5.4 event) and recent extreme rainfall (2024–2025)—will try to highlight relationships between antecedent moisture conditions and seismic history to define slope stability. This integrated analysis is expected to provide fundamental insights into event timing, shaking intensity, and the ultimate magnitude of landslide movements. Ultimately, the project will offer a robust, multi-sensor framework for multi-hazard risk assessment in complex terrain.
How to cite: Grezio, A., Fornaciai, A., Devoti, R., Borghi, A., Burrato, P., Delrosso, D., Di Pietro, P., Favali, M., Ghirotti, M., Nannipieri, L., Perrone, L., Piacentini, T., Piana Agostinetti, N., Pintori, F., and Tarabusi, G.: Integrating Remote Sensing and Ground Observations to Discriminate Landslide Induction and Understand Trigger Interactions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12083, https://doi.org/10.5194/egusphere-egu26-12083, 2026.
Rocky coasts are highly dynamic environments in which the interaction of multiple natural factors controls landscape evolution and the generation of natural hazards. Along the Cantabrian coast of northern Spain, particularly in the Asturian sector, an increase in gravity-driven erosion processes affecting coastal cliffs has been observed in recent years (Domínguez-Cuesta et al., 2022a, 2022b). One of the most representative examples is the large landslide affecting the cliff at the Tazones Lighthouse (Asturias). This landslide has been the subject of historical observations and a systematic monitoring programme, which has allowed its behaviour to be analysed over medium and long-term timescales. This contribution presents data collected over eight years of monitoring and highlights the scientific value of long-term monitoring programmes for improving the understanding of coastal landslide dynamics and their relationship with cliff retreat.
References:
Domínguez-Cuesta, M.J., González-Pumariega, P., Valenzuela, P., López-Fernández, C., Rodríguez-Rodríguez, L., Ballesteros, D., Mora, M., Meléndez., M., Marigil, M.A., Pando, L., Cuervas-Mons, J. y Jiménez Sánchez, M. (2022a). Marine Geology 449, 106836.
Domínguez-Cuesta, M.J., Rodríguez-Rodríguez, L., López-Fernández, C., Pando, L., Cuervas-Mons, J., Olona, J., González-Pumariega, P., Serrano, J., Valenzuela, P. y Jiménez Sánchez, M. (2022b). Remote Sensing, 14(20), 5139.
Research funding: RETROCLIFF (PID2021-122472NB-100, MCIN/AEI/FEDER, UE) and GEOCANTABRICA (IDE/2024/000753, SEK-25-GRU-GIC-24-072, Principado de Asturias).
How to cite: Carrillo Bravo, J. A., Domínguez Cuesta, M. J., González-Pumariega, P., Cuervas-Mons, J., Rodríguez-Rodríguez, L., Mateos, F., López-Fernández, C., Pando, L., Valenzuela, P., and Jiménez-Sánchez, M.: Understanding coastal landslide dynamics through long-term monitoring: the Tazones Lighthouse case study (N Spain), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18965, https://doi.org/10.5194/egusphere-egu26-18965, 2026.
Mountain highway networks, characterized by low-redundancy topologies and critical routes traversing geologically active zones, are highly vulnerable to cascading failures triggered by landslides, which can severely disrupt regional connectivity. Currently, there is a lack of universal and efficient methods for accurately identifying, at a regional scale, high-risk landslides that pose substantial threats to highway infrastructure among numerous detected instabilities. To address this gap, this study proposes an integrated pre-disaster risk assessment framework combining interferometric synthetic aperture radar (InSAR), the vector inclination method (VIM), the sloping local base level method (SLBL), and empirical models for the systematic identification and risk prioritization of unstable highway landslides. The framework consists of four core components: wide-area landslide detection, multi-dimensional displacement reconstruction, accurate volume inversion, and run-out distance prediction. Applied in the Jishishan region, the framework identified 530 potential landslides using small baseline subset InSAR (SBAS-InSAR) technology. Among these, 197 were initially determined to potentially threaten highways. By integrating VIM and SLBL methods, landslide volumes were reliably estimated, ranging from 8 × 10³ m³ to 3.3 × 10⁸ m³. Furthermore, six empirical models were employed to rapidly predict potential run-out distances based on landslide volume and topographic parameters, yielding results between 24 m and 2460 m. By comparing these predicted run-out distances with the actual distances to highways, 113 landslides were confirmed to pose realistic threats. Additionally, complex network theory was introduced to evaluate the impact of landslide-induced highway interruptions on regional connectivity. The results show that approximately 17.75% of highway sections in the region fall into "major" or "critical" importance categories, while about 32.74% of the landslides exhibit "major" or "critical" network disruption potential. The failure of such landslides would significantly impair regional transportation functionality, necessitating prioritized risk mitigation and engineering interventions. The proposed non-contact, wide-area applicable risk assessment framework, which provides a scientific basis for precise risk prevention and control in highway systems, is particularly suitable for topographically complex and inaccessible mountainous areas, thereby supporting optimal allocation of disaster mitigation resources.
How to cite: Fan, Q., Lu, Z., and Zhao, J.: From identification to prioritization: A comprehensive framework for assessing landslide risks to mountain highway networks combining InSAR-derived displacements, volume estimation, and run-out prediction, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8514, https://doi.org/10.5194/egusphere-egu26-8514, 2026.
Frozen Debris Lobes (FDLs) are slow-moving, permafrost-related landslides affecting hillslopes in the Brooks Range (Alaska). Their recent acceleration, driven by climate warming, is increasingly relevant for the long-term management of key Arctic infrastructure, notably the Dalton Highway—the only road access to the North Slope oil fields—and the trans-Alaska pipeline. FDL activity already has required operational responses within this corridor, including the realignment of the Dalton Highway in 2018 to provide more distance from FDL-A. While detailed field monitoring (RTK-GPS surveys and in situ instrumentation) has been conducted on a limited number of lobes since 2012, corridor-scale assessment of FDL dynamics increasingly relies on remote sensing; however, optical observations in the Arctic are hindered by frequent cloud cover, and validation datasets remain scarce at high latitudes. Here we quantify ground-surface displacements for nine FDLs using Ground Deformation Monitoring with OPTical image time series (GDM-OPT-SLIDE; DATA-TERRA/FormaTerre), an automated processing chain that extracts horizontal surface displacements from Sentinel-2 image time series. We validate satellite-derived displacement rates using a unique GPS dataset collected since 2012. The resulting rates agree with ground observations (R² = 0.71) and reveal a marked acceleration in 2020 followed by a slowdown from 2022 onwards. Because FDL surfaces exhibit heterogeneous land cover (trees, shrubs, and bare soil), we assess land-cover effects using high-resolution LiDAR data. Results indicate that agreement with ground observations improves under dense forest cover. Overall, this study highlights the potential of optical satellite monitoring to track periglacial slope dynamics in warming Arctic permafrost terrain, enabling systematic regional mapping of landform displacement and supporting investigation of climatic controls at the regional scale.
How to cite: Bayle, A., Darrow, M. M., Corona, C., Provost, F., Michéa, D., Malet, J.-P., and Stoffel, M.: Tracking Frozen Debris Lobe ground deformation in the Brooks Range (Alaska) using Sentinel-2 optical image time series validated by long-term GPS, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21669, https://doi.org/10.5194/egusphere-egu26-21669, 2026.
Slovenia is currently completing the country-wide mapping of landslide, rockfall, and debris flow susceptibility at the scale of 1:25,000. As outlined in the EGU contribution (EGU25-8639, https://doi.org/10.5194/egusphere-egu25-8639) the concept relies on fuzzy logic and linear membership functions, attributing weights to relevant factors for the formation of landslides, rock falls, and debris flows. As a static susceptibility (i.e., not probability) concept, the notion of return periods for events of specific characteristics does not apply.
The three susceptibility models rely on pixel-wise calculated algorithms that draw on different sets of factors: 6 factors for the landslide, 4 factors for the rockfall, and 7 factors for the debris flow susceptibility model.
One of the factors processed for the landslide as well as the rockfall susceptibility model is the interplay of the aspect of the slope and the azimuth to the dip direction of the underlying lithology in a distinct pixel. Named shortly “aspect” and “dip direction”, both angles are measured horizontally; thus, the factor is named horizontal synchronism.
In analogy, also the vertical synchronism should be possible to compute between the slope and the dip in the same pixel; however, this computation seems not as straightforward to implement since the question of the reference plane arises.
In this contribution, we outline the currently implemented horizontal synchronism model between aspect and dip direction and point out its strengths and weaknesses. Moreover, we move from a possible 2D approach for the vertical synchronism model between slope and dip to a possible formulation in 3D.
How to cite: Domej, G., Jež, J., Novak, A., Milanič, B., Markelj, A., and Praprotnik Kastelic, J.: Landslide Susceptibility Maps in Slovenia: How to account for the interplay of slope and dip?, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3921, https://doi.org/10.5194/egusphere-egu26-3921, 2026.
In recent years, frequent heavy rainfall and strong earthquakes have caused numerous landslides, debris flows, and barrier lake disasters in Taiwan. This study applies terrain analysis on real landslide barrier lakes to develop an interpretation method for the potential location of barrier lake induced by a massive landslide. This study applies GIS and a 1 m-resolution DEM to analyze landslide cases. First, a microtopographic feature database for a massive landslide is established, from which types of large landslide masses are identified. Next, the length and width boundaries of the landslide body are determined. The SLBL sliding-surface estimation method proposed by Jaboyedoff et al. (2019) is then applied, assuming the landslide scarp and the outcrop point as the start and end boundaries of the sliding surface, and using a quadratic parabolic equation to infer the intermediate failure surface. Finally, the cross-sectional areas of the segmented transverse profiles along the longitudinal sliding surface are calculated to estimate the landslide volume. After obtaining the landslide volume, the equivalent friction method is used to estimate the horizontal transportation distance of the landslide mass, in order to assess whether the landslide debris can reach the opposite river bank and form a landslide dam that blocks the river. Subsequently, the model proposed by Chen et al. (2014) is applied to infer the dam length. Together with the dam length, water depth, and channel width are used to calculate the minimum blocking volume. Last, by comparing the actual landslide volume with the minimum blocking volume, it is then determined whether a landslide lake can be formed. 25 landslide barrier lake events in the Gaoping River Basin, Taiwan, were used to test the interpretation method for potential locations. This interpretation method can serve as a predictive tool for identifying potential landslide barrier lake sites, thereby reducing the impact of disasters.
How to cite: Chen, T.-C., Chang, W.-C., and Chen, K.-T.: Geomorphological Interpretation Method on the Potential Location of Barrier Lake Induced by Massive Landslide, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4453, https://doi.org/10.5194/egusphere-egu26-4453, 2026.
Landslide dam hazards have become one of the disaster types of increasing global concern. Once a landslide dam is formed, its typically short lifespan, high uncertainty, and sudden failure can pose severe threats to protected targets both upstream and downstream. The ability to identify locations within landslide-prone areas where landslide dams are likely to form would substantially enhance disaster prevention and mitigation capabilities. This study focuses on historical landslide events and landslide-dammed lake cases. Satellite imagery is employed to quantify key geomorphic parameters, including landslide elevation, slope gradient, and channel width. Both the longitudinal accumulation characteristics along the river channel and the lateral mobility of the landslide mass are considered. Through nondimensional analysis, this study seeks to identify quantitative threshold values linking historical landslide characteristics to landslide dam formation. These thresholds are used to preliminarily assess the likelihood of landslide dam formation in potential landslide hazard areas and to further delineate locations with dam-formation potential. The results are expected to provide a scientific basis for improving early warning systems and disaster prevention and mitigation planning related to landslide dams.
How to cite: Chen, K.-T., Tsai, H.-J., and Chen, T.-C.: Geomorphic Characterization and Quantitative Assessment of Landslide Dam Formation in Potential Landslide Hazard Areas, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4750, https://doi.org/10.5194/egusphere-egu26-4750, 2026.
Landslides remain a major natural hazard with persistent gaps in global and regional inventories, largely due to the high cost and effort of field-based documentation. To address this, we investigate automated, minimal-input approaches for historical landslide detection using Sentinel-2 NDVI time series (2018–2024) across Bavaria, Germany. The study introduces the Independent Baseline Method (IBM), a novel unsupervised framework leveraging external, landslide-free reference data to mitigate baseline contamination, and compares it with two adapted techniques—Statistical Window Analysis (SWA) and Seasonal-Trend Decomposition (STL).
Evaluation across 15 documented landslide events shows that IBM delivers the most balanced and robust performance. While SWA yielded higher sensitivity, it also generated extensive false positives, whereas STL showed limited detection capacity due to baseline distortion. Detection success was positively correlated with landslide size, confirming the scalability of the approach for medium to large events. A systematic analysis also identified errors in Sentinel-2’s Scene Classification Layer as a dominant source of false detections, primarily linked to atmospheric misclassification.
Despite such constraints, IBM successfully identified previously undocumented landslide occurrences, subsequently confirmed through inventory updates. These results demonstrate that NDVI-based, low-complexity frameworks can meaningfully enhance the completeness of landslide records. The proposed approach, relying solely on open-access EO data and minimal reference information, establishes a scalable, transferable, and cost-efficient foundation for regional landslide monitoring. It also illustrates how strategic use of external baselines can substantially improve unsupervised change detection performance, paving the way for operational applications in risk assessment and environmental management.
How to cite: Geiß, C., Happ, S., Sapena, M., Aravena Pelizari, P., and Taubenböck, H.: Unsupervised Landslide Detection Using Multitemporal Sentinel-2 Imagery, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7530, https://doi.org/10.5194/egusphere-egu26-7530, 2026.
The fusion of UAV-based LiDAR and RGB surveys with geotechnical, geophysical, and hydrogeological field investigations enables a detailed characterization of the Cuejdel Lake landslide dam and its host landslide. To identify the main landslide features and to map the morphology of the study area, a 3D point cloud was generated from a UAV-based LiDAR survey covering an area of 126 ha on the western slope of the Muncelu Peak. Based on this dataset, a 3D surface model was constructed and textured using RGB imagery from a separate UAV-based photogrammetric survey, revealing the spatial distribution and characteristics of sedimentary facies within the eroded spillway outcrop.
A two-dimensional plane representing the results of an electrical resistivity tomography (ERT) survey across the landslide dam was integrated into the 3D model, allowing sedimentary facies to be linked to distinct resistivity zones. The orientation of intact stratigraphy measured in the field was incorporated and extrapolated until intersecting the ERT plane. In addition, representative facies were sampled for grain-size analysis. Thirteen infiltration tests conducted parallel to the ERT profile provided proxy permeability values that were also integrated into the model.
The investigations reveal that the feature previously interpreted as a single landslide actually consists of two distinct landslides, of which the northern landslide impounded Cuejdel Lake. Facies mapping shows a highly heterogeneous structure composed of large intact flysch blocks embedded in a low-permeability matrix of sand and clayey silt. Despite this heterogeneity, infiltration measurements indicate a relatively uniform permeability within the saturated phreatic zone, with values between 1 × 10⁻⁷ and 1 × 10⁻⁸ m s⁻¹. While facies distributions, laboratory analyses, and resistivity patterns indicate strong internal heterogeneity, the hydraulic behavior of the dam is controlled by the mixture of sand, silt and clay.
This comparatively impermeable structure facilitated rapid lake-level rise and temporary overtopping during the early stage of dam formation. However, geomorphic evidence, water marks on tree trunks, and historical records indicate that this initial overtopping phase was halted by winter-induced lake-level lowering, after which erosion shifted to progressive spillway incision at the landslide toe during the following season.
How to cite: Wolfert, T., Mihu-Pintilie, A., Dufresne, A., Stoleriu, C., Trifanov, C., and Amann, F.: UAV-based characterization of the Cuejdel Lake landslide dam (Romania) integrating LiDAR, photogrammetry, geophysics and hydrogeological properties, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10371, https://doi.org/10.5194/egusphere-egu26-10371, 2026.
Interferometric Synthetic Aperture Radar (InSAR) deformation monitoring in vegetated mountainous areas often faces challenges, including temporal decorrelation and atmospheric noise. Extracting true surface deformation trends from time-series signals and validating their accuracy remain critical issues. This study focuses on the Lantai area in Northeast Taiwan, utilizing Sentinel-1 satellite imagery from 2021 to 2025. We employed the Small Baseline Subset (SBAS-InSAR) technique to resolve surface deformation, with a specific focus on optimizing time-series signals and performing validation analysis using in-situ GNSS data.
To optimize the time-series deformation signals, this study compared the efficacy of three smoothing methods on the raw SBAS results: Mean Filter, Median Filter, and Gaussian Filter. The results indicate that while the Mean Filter is computationally efficient, it tends to cause boundary blurring and time delays. The Median Filter effectively removes sudden noise spikes but performs less effectively in smoothing subtle continuous changes. In contrast, the Gaussian Filter successfully suppresses noise while preserving waveform continuity, making it the most suitable method for analyzing long-term deformation trends in this study area.
Regarding accuracy validation, the study compared the optimized InSAR time-series deformation with data from continuous GNSS monitoring stations. The comparison reveals that, due to the 12-day satellite revisit cycle and dense vegetation, the InSAR results exhibit a noticeable short-term drift effect. However, over the five-year observation period, the overall cumulative deformation trends between InSAR and GNSS show good consistency. This research confirms that with appropriate filter optimization, Sentinel-1 time-series InSAR technology can be effectively applied to broad-area surface deformation screening in mountainous regions, providing reliable long-term trend data for landslide potential zoning.
How to cite: Wang, K.-L., Hsu, Y.-J., Lin, M.-L., Chen, R.-F., and Chan, K.-C.: Optimization of Sentinel-1 Time-Series InSAR Deformation Signals and GNSS Validation Analysis, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6089, https://doi.org/10.5194/egusphere-egu26-6089, 2026.
Brazil has experienced an increase in landslides occurrences associated with extreme rainfall events. In the Taquari-Antas Basin, southern Brazil, the valley-shaped relief favor the development of different types of landslides. During the 2024 extreme rainfall-induced event, more than 16,000 landslide scars were mapped across the state, including 281 in the municipality of Santa Tereza, which presents the highest landslide scar density per area. However, landslides inventories are still largely based on visual interpretation of satellite imagery, manual delimitation, and, when feasible, field validation.
Satellite imagery plays a fundamental role in landslide mapping, particularly in hard-to-reach areas and during disaster events. To improve landslide detection, this study proposes a simple approach combining the variation of Normalized Difference Vegetation Index (dNDVI) and terrain slope, using Sentinel-2 imagery with 10 m spatial resolution. Data processing was performed using Google Earth Engine (GEE).
The dNDVI, calculated from pre- and post-event images, enables the identification of vegetation loss, which is particularly effective in Santa Tereza, where more than 70% of the municipality is forest-covered. As landslides predominantly occur on steep hillslopes in Brazil, slope information was incorporated to refine the detection. The combined analysis of dNDVI and slope resulted in an initial landslide detection map.
NDVI values range from 0 to 1, with higher values indicating denser vegetation. In southern Brazil, low dNDVI thresholds (e.g., 0.10) may misclassify cloud shadows or crop harvesting as landslides, whereas high thresholds (e.g., 0.40) may capture only the core of the scar. A sensitivity analysis was conducted by testing three dNDVI thresholds (0.25, 0.20, and 0.15) combined with three slope thresholds (15°, 10°, and 8°).
Validation was performed by comparing the detection results with a detailed landslide inventory produced by the Latitude/UFRGS research group, classifying the outcomes as true positives, false negatives, and false positives.
The results show true positive rates ranging from 59% to 83%. The best overall performances were the combinations dNDVI ≥ 0.15 with slope ≥ 15°, dNDVI ≥ 0.20 with slope ≥ 15°, and dNDVI ≥ 0.25 with slope ≥ 15°. False negative rates were lowest for dNDVI ≥ 0.15 with slope ≥ 15° combination. False positive rates ranged from 72% to 87%, with lower values observed for combinations of higher dNDVI and slope thresholds. The proposed approach provides a practical and rapid technique to support landslides mapping and post-disaster monitoring.
Acknowledgements: This study was supported by FAPERGS under Grant Agreement No. 24/2551-0002124-8 (Call FAPERGS 06/2024) and No. 25/2551-0002522-2 (Call FAPERGS 05/2025).
How to cite: Andrades Paixão, M., Lahiguera Cesa, L., Fossa Sampaio Mexias, L., and de Oliveira Andrades-Filho, C.: Mapping landslides using NDVI variation and slope in Google Earth Engine: A case of study of Santa Tereza municipality due to the most extensive disaster in Brazil (2024), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14916, https://doi.org/10.5194/egusphere-egu26-14916, 2026.
Landslide is one of the most commonly happened and threatening natural hazards in Taiwan. Because of the complicated terrain, geological, geotechnical and weather conditions landslides are frequently triggered by earthquakes, typhoons or heavy rainfalls in Taiwan, and sometimes result in serious damages. Satellite imagery is one of the commonly used sources to determine the extent of landslides for mapping, inventorying, assessment and hazard mitigation decision support. However, conventional image-based landslide detection approaches rely only on spectral and two-dimensional spatial characteristics, which may not be able to achieve high accuracy and difficult to differentiate different landslide-related terrain interpretations. This research integrates topographic features with high resolution satellite images to improve landslide detection and monitoring effectiveness and efficiency. In addition to just including height information from three-dimensional (3D) point clouds or digital elevation/terrain models (DEM/DTM), multi-scale landslide-related topographic features are derived from 3D DTM generated from airborne LiDAR surveys or stereo satellite images. These features include: slope, curvature, surface roughness, topographic position index, geomorphons and geo-hydrological features etc. These features are essential for identifying important landslide terrains, such as hummocky, crown scarps, toe bulges, gullies and the like. Based on the calculated topographic features, landslide candidate areas can be identified according to a developed scoring/classification equation. The derived topographic features and resultant landslide candidate (scores) are integrated with high resolution satellite images for landslide detection. A deep learning model based on ResUet is utilized to identify landslide areas. The developed framework was applied to analyze multi-temporal satellite images and digital terrain data of a mountainous watershed region in southern Taiwan. Preliminary results indicate that integrating topographic features with satellite images can improve the performance of landslide detection and is an effective approach for long-term monitoring of large- areas vulnerable to landslide hazards.
How to cite: Tsai, F., Chen, W., and Lo, C.-C.: Integrating Topographic Features and Satellite Images with Deep Learning for Landslide Mapping and Monitoring, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15719, https://doi.org/10.5194/egusphere-egu26-15719, 2026.
This study presents a methodology for monitoring slow-moving landslides across Central Nepal (21,500 km²) using multi-temporal InSAR analysis. The region is characterized by complex terrain and dynamic environmental conditions, posing challenges related to the large volume of SAR data, diverse acquisition modes, steep Himalayan topography, dense vegetation, seasonal variations, and high ground displacement velocities.
To address these challenges, a fully automated, computationally optimized, and self-evaluating processing chain was developed using the AMSTer Toolbox (Derauw et al., 2020; d'Oreye et al., 2021; Smittarello et al., 2022). The chain processes Sentinel-1 archives across five orbital tracks (Ascending 85 and 158; Descending 19, 92, and 121), comprising approximately 1,500 images and generating over 4,500 interferometric pairs. The system is also capable of handling ERS, ENVISAT, TSX, PAZ, and ALOS data for smaller portions of the region, and is prepared for the upcoming NISAR L-Band mission. Deformation maps are inverted using the MSBAS method (Samsonov and d'Oreye, 2012) to extract mean linear velocity maps and time series in Line of Sight and/or vertical and horizontal components.
This study involves evaluating the impact of baseline selection criteria on displacement measurement accuracy. Using the Gayu Kharka landslide (Mustang region) as a calibration site, where optical imagery (Planet, Pléiades) indicates velocities of 12-15 cm/yr westward and 18-22 cm/yr southward, different temporal baseline strategies were systematically compared. Connecting each image to only the 1-3 shortest temporal neighbors provides best velocity estimates.Adding pairs with longer temporal baseline configurations (Bt 100- 400 days) fails to capture rapid movements by introducing phase aliasing.
The effectiveness of Sentinel-1 ETAD (Extended Time Annotation Products) corrections for ionospheric, tropospheric, and geodetic effects was also assessed. Preliminary results indicate ETAD reduces vertical displacement standard deviations by factors of 2-3 under favorable conditions, though performance varies depending on atmospheric state.
Additionally, 3D velocity decomposition using the Surface-Parallel Flow constraint was explored, enabling extraction of North-South displacement components. Initial results from the Bolde landslide, compared against continuous GNSS measurements from a newly installed network, demonstrate the method's capability to resolve three-dimensional displacement patterns.
How to cite: Glaude, Q., d'Oreye, N., Smittarello, D., Derauw, D., Jaspard, M., Barrière, J., Samsonov, S., Celli, G., and Maury, L.: Monitoring Slow-Moving Landslides in Central Nepal using Multidimensional Small Baseline Subset (MSBAS) with the AMSTer Toolbox, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14760, https://doi.org/10.5194/egusphere-egu26-14760, 2026.
Sinkholes are a major type of geological hazard worldwide, commonly formed through collapse processes driven by hydrological dynamics. In the Loess Plateau region, sinkholes represent a distinctive loess subsurface erosion–related hazard and are widely distributed near tableland margins and slope areas. Once sinkholes reach a certain scale, they can significantly reduce slope stability, trigger cascading hazards, and threaten infrastructure such as roads, pipelines, industrial facilities, and residential buildings. Consequently, accurate identification of loess sinkholes is essential for disaster prevention and mitigation in loess regions. Traditional sinkhole identification relies mainly on field investigations, which are time-consuming and labor-intensive when applied at large scales. In recent years, high-resolution topographic data combined with machine learning techniques have been increasingly used for sinkhole detection. Approaches based on light detection and ranging (LiDAR)-derived digital elevation models (DEMs), including contour-based methods, random forests, and deep learning models trained on elevation, slope, and shaded relief images, have shown promising results, particularly for large-scale karst sinkholes with pronounced topographic relief. However, loess sinkholes are typically small in size and characterized by subtle micro-relief, making them difficult to distinguish in DEM imagery. To address this challenge, some studies have integrated unmanned aerial vehicle (UAV) thermal imagery with machine learning methods, while others have applied modified U-Net architectures with multi-scale filtering to improve identification accuracy. Recent investigations have also explored the use of unmanned aerial systems, handheld laser scanners, and point cloud learning networks such as PointNet++ for loess sinkhole detection. Despite their effectiveness, these methods are limited by high equipment costs, field survey constraints, and safety concerns. Moreover, the unique physical properties of loess and the distinct size, morphology, and spatial distribution of loess sinkholes further complicate their identification, leading to limited performance of existing methods. To overcome these limitations, this study employs wavelet transforms to decompose sinkhole data into multi-frequency components for enhanced feature learning. In addition, a Kolmogorov–Arnold network is introduced to strengthen nonlinear boundary representation. Experimental results demonstrate that the proposed method achieves high accuracy, efficiency, and strong generalization across multiple loess regions.
How to cite: Pang, Z. and Zhu, W.: Identification of Loess Sinkholes Using a Gaussian Radial Basis Kolmogorov–Arnold Network, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3323, https://doi.org/10.5194/egusphere-egu26-3323, 2026.
Posters virtual: Fri, 8 May, 14:00–18:00 | vPoster spot 3
EGU26-11366 | ECS | Posters virtual | VPS14
Transferability of Semi-Automatic Landslide Mapping Approach Using High-Resolution DTMs: a Case Study from the Swabian Alb, GermanyFri, 08 May, 14:18–14:21 (CEST) vPoster spot 3
Landslides are among the natural hazards that significantly impact human life and infrastructure, making accurate landslide mapping essential for hazard assessment, risk reduction, and land use planning. However, mapping landslides, particularly in vegetated areas, remains challenging, as traditional field-based and manual mapping approaches are time-consuming and require substantial expert knowledge. Semi-automatic mapping methods based on high-resolution Digital Terrain Models (DTMs) have improved landslide inventory preparation; however, their transferability to larger and diverse environmental settings remains limited and require further assessment. Therefore, this study aims to assess the transferability of a Geographic Object-Based Image Analysis (GEOBIA) landslide mapping approach using optimal moving window sizes, and to examine whether model performance varies across specific land use classes and improves with higher-quality DTM data.
A GEOBIA-based model, originally developed for forest covered landslides in the cuesta landscape of Jena region (Zangana et al., 2025), was transferred and applied to landslides at the Swabian Alb escarpment in south-western Germany, which are located not only in forests, but also in grasslands and settlements. The study area is characterized by Jurassic limestones overlying marls and clays. It is affected mainly by rotational slides, slump-earthflows, and translational landslides, some of which show repeated reactivation. The manually mapped landslide inventory was used for result validation and accuracy assessment. DTM derivatives (from the 2003 and 2023 data) were prepared using optimal moving window sizes following Zangana et al. (2025). The semi-automatic landslide detection workflow involved multi-resolution segmentation (MRS) and support vector machine (SVM) classification, followed by expert-based refinement and accuracy assessment against the reference map. Finally, transferability was further examined through land use class-based performance analysis and by evaluating the effect of higher-quality 2023 DTM data on model results.
The results indicate that the model developed for the Jena region is transferable to the Swabian Alb. When applied to the 2003 dataset, without differentiating between land use types, the model achieved a 75% detection rate for landslide body areas. Using the 2023 dataset increased detection accuracy to 86% compared to the 2003 data. The area-based detection accuracy in this study is approximately 30% higher than reported for the Jena region by Zangana et al. (2025). When considering only forested areas—for which the model was originally developed—the true positive rate increased by about 15%, while false positives decreased by a similar margin. Although the approach effectively identifies landslides, particularly in vegetated areas, it currently performs best for cuesta-related rotational slides. Further assessment and refinement are needed to extend its applicability to other landslide types. Nevertheless, the method shows strong potential for detecting landslides with distinct geomorphological signatures in high-resolution DTMs worldwide.
Reference: Zangana, I., Bell, R., Drăguţ, L., Sîrbu, F., and Schrott, L.: Mapping forest-covered landslides using Geographic Object-Based Image Analysis ( GEOBIA ), Jena region , Germany, Nat. Hazards Earth Syst. Sci., 25, 4787–4806, https://doi.org/10.5194/nhess-25-4787-2025, 2025.
How to cite: Zangana, I., Bell, R., Drăguţ, L., and Schrott, L.: Transferability of Semi-Automatic Landslide Mapping Approach Using High-Resolution DTMs: a Case Study from the Swabian Alb, Germany, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11366, https://doi.org/10.5194/egusphere-egu26-11366, 2026.
