PICO: Mon, 28 Apr | PICO spot 3
Landslide inventories play a crucial role in assessing susceptibility, hazards, and risks, particularly in mountainous regions where devising resilience strategies becomes essential. The significance of such inventories becomes even more pronounced in the context of climate change, which may render existing databases inadequate due to evolving stability conditions. A clear illustration of this was seen in May 2023, when the Emilia-Romagna region in Italy experienced two significant rainfall events. These events triggered widespread flooding and thousands of landslides, including shallow debris slides and flows on slopes that had been considered stable, as historical data had not recorded previous landslides there. The total damages have been estimated to surpass 9 billion euros, affecting roads, railways, buildings, and cultural heritage sites, along with the destruction of bridges, power facilities, and communication lines. Additionally, agricultural fields, farming operations, and cultivated slopes saw significant disruption over an area of about 1000 km². Fifteen people lost their lives due to the flooding and two due to landslides.
In the aftermath, our team supported local and national agencies by engaging in field surveys and immediate assessments to address urgent public safety concerns. Our focus then shifted to mapping the landslides, initially identifying impacted roads and buildings to coordinate emergency responses and perform preliminary damage evaluations. We subsequently completed a detailed landslide inventory, producing a comprehensive map of all landslides triggered by the rainfall. This map has now been adopted by the Po River Authority and the Emilia-Romagna region as the official record for the May 2023 event and is being used by the Commission for Reconstruction to guide the recovery efforts.
The landslide inventory includes 80,997 polygons and has been made publicly available through the Zenodo repository (DOI: 10.5281/zenodo.13742643; https://essd.copernicus.org/preprints/essd-2024-407/). The dataset is provided in a shapefile format, which includes detailed attributes like the type of landslide and the geological unit of each polygon, facilitating in-depth analysis. Additionally, the Emilia-Romagna region's geoportal offers unrestricted access to extensive spatial data, which can be integrated with our landslide map to refine both traditional and advanced machine-learning predictive models. The frequent shallow planar failures observed during the event also offer an exceptional opportunity to test physically-based slope stability models. We invite the scientific community to utilize this dataset or to collaborate on research projects that could leverage this tragic event to deepen our understanding of landslide risks.
How to cite: Berti, M., Pizziolo, M., Scaroni, M., Generali, M., Critelli, V., Mulas, M., Tondo, M., Lelli, F., Fabbiani, C., Ronchetti, F., Ciccarese, G., Dal Seno, N., Ioriatti, E., Rani, R., Zuccarini, A., Simonelli, T., and Corsini, A.: RER2023: a freely accessible landslide inventory dataset from the May 2023 Emilia-Romagna event, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5648, https://doi.org/10.5194/egusphere-egu25-5648, 2025.
Shallow landslides are among the most frequent and impactful geomorphological phenomena in areas affected by intense rainfall and complex lithological settings. In this study we present and analyse a new multi-temporal shallow landslide inventory for the Apuan Alps area (Northern Tuscany, Italy). The study area, covering 625 km², is characterized by high landslide susceptibility due to both occurrence of intense rainfall events and complex morphology and structural geology conditions. A visual interpretation of high spatial resolution orthophotos was performed to recognize and map both landslide features and examples of stable areas. The aerial images used for landslide mapping cover a period of 67 years, from 1954 to 2021. The acquisition was not evenly distributed over time, with intervals between successive images varying from 2 to 24 years and averaging approximately 6 years. Nevertheless, the last two decades (2003-2021) saw a more consistent acquisition rate, with aerial images captured every 3 years. The dataset, made up of 1433 positive landslide entities and 100 stable areas, was validated through field surveys, achieving an overall accuracy of 91%. During field validation, further information were acquired, such as movement type, material involved and scarp height. The overall inventory underwent spatial, temporal and statistical analysis. Spatial analysis revealed two high-density clusters (>25 landslides/km²), primarily associated with extreme rainfall events occurred in 1996, 2010, and 2012. Temporal analysis highlighted a significant increase of normalized annual landslide frequency during the recent decades; also the relationships with the increase of intense rainfall events was explored. Magnitude–frequency distribution analysis exhibited a negative power-law relationship for medium and larger landslides, with a rollover at areas around 100 m². The shape and the parameters of the magnitude-frequency relationship well fit to other functions published in the literature. In a general perspective, the new inventory shows high frequency of “small” landslides, which instead are almost lacking within published landslide inventories for the study area (e.g. IFFI, Inventario Fenomeni Franosi Italiani). The intersection between shallow landslides and bedrock lithological units allowed us to recognize highest landslide density for silt and clay-rich lithologies, such as flysch and metarenites, while carbonate units is characterized by higher stability. Analysis of morphometric variables revealed that south- and southeast-facing concave hillslopes with gradients between 30° and 50° are particularly susceptible to landslides. These results align with previous research highlighting the role of slope aspect, steepness, and contributing area in landslide initiation. Field validation provided further insights into the dynamics and geometry of shallow landslides. Avalanches were the most common type (50%), followed by slides (30%), flows (15%), and falls (5%). The variability for scarp height and sliding surface location highlights the involvement of both slope deposits and bedrock, providing relevant clues to help both understanding failure mechanisms and improving approaches for susceptibility models. This research highlights the added value of integrating remote sensing – based data extraction, spatial analyses, field validation and statistical methods to enhance the understanding of shallow landslide processes. Moreover, the inventory represents a new robust, high-quality dataset suitable for landslide susceptibility and hazard zoning.
How to cite: D'Addario, E., Masoni, G., Marques e Silva Rocha de Oliveira, E., Pippi, M., and Disperati, L.: Shallow landslides in Northern Tuscany (Italy): a new multi-temporal inventory and its spatial and statistical analysis, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19624, https://doi.org/10.5194/egusphere-egu25-19624, 2025.
Landslide inventories are critical to support investigations of where and when landslides have happened and may occur in the future. They can be developed using different techniques and data, each bringing intrinsic limitations and potential sources of mapping errors, hence affecting the overall accuracy and reliability of the subsequent analyses.
For more than one decade, the Geotechnical Engineering Group (GEG) of the University of Salerno (Italy) has been carrying out a specific research activity aimed at collecting and organizing, within a national landslide catalogue called “FraneItalia”, information on landslides that occur in Italy from online news sources. To this aim, the news aggregator Google Alerts has been used for screening web pages and news articles published in Italian language. The FraneItalia catalogue is freely accessible at https://zenodo.org/records/7923683. A description of the main features of the catalogue and the procedures adopted to fill it out can be found at https://doi.org/10.1186/s40677-018-0105-5.
FraneItalia, which is being continuously updated, to date contains data on more than 9000 landslide events that occurred in Italy during the period 2010-2024. The catalogue includes both fatal landslide events and events that did not produce physical harm to people. The main peculiarity of the catalogue is the distinction between single landslide events, SLE (i.e., records only reporting one landslide) and areal landslide events, ALE (i.e., records referring to multiple landslides triggered by the same cause in the same geographic area). The structure is organized as a database where each reported landslide event is characterized by 40 unique fields, which are grouped in 9 thematic tables: main info; spatial information; temporal information; landslide characteristics; consequences to people, structures, infrastructures, cars and other elements; and source. Not all fields are mandatory. A set of constraints has been adopted to ensure the correctness and the semantic integrity of the attributes. In addition, a set of confidence descriptors are associated to each landslide record to measure the level of accuracy of the spatial and temporal information. Indeed, the availability of accurate and up-to-date information is essential for improving the accuracy and the quality of the subsequent analyses in landslide research.
Different subsets of the catalogue have been already used to carry out studies on landslide risk in Italy (https://franeitalia.wordpress.com/publications/), including: calibration and validation of models for landslides prediction at territorial scale; detection and mapping of spatio-temporal clusters of landslides; susceptibility, hazard, and risk assessment. Given the rising demand for high-quality data to be used in comprehensive analyses at regional and national scales, this dataset might be very useful for supporting decision-making in landslide risk management in Italy. Moreover, the methodology to define and populate FraneItalia is deemed to be general and can be used to develop similar initiatives in other countries.
How to cite: Pecoraro, G. and Calvello, M.: Potentials and limitations of a landslide inventory based on online news sources: the FraneItalia catalogue, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5768, https://doi.org/10.5194/egusphere-egu25-5768, 2025.
Landslides are geomorphological hazards triggered by natural factors such as rainfall, seismic activity, and snowmelt. A landslide inventory is essential for understanding and assessing the processes, distribution, and risks associated with these events. While traditional manual mapping from aerial imagery delivers high accuracy for slope-scale studies, it is labor-intensive and impractical for large-scale applications. Advances in remote sensing technologies, including high-resolution satellite imagery and synthetic aperture radar (SAR), have significantly enhanced the efficiency of landslide detection over broader geographical scales. Concurrently, deep learning techniques, such as convolutional neural networks (CNNs), have revolutionized the field by automating landslide feature extraction and segmentation through remote sensing imagery, addressing the inefficiencies of manual methods.
Current challenges include the regional distribution bias in remote sensing-based landslide datasets, with limited high-quality data available for regions like Japan. Additionally, the lack of systematic evaluation of optimal features and deep learning architectures hinders improvements in detection accuracy and model transferability.
To address these gaps, we developed and validated the Japan High-Resolution Landslide Dataset (JHRLD), which integrates multi-sensor data encompassing spectral, SAR, and topographic features. The dataset comprises two subsets: Sentinel-2 for moderate-resolution (10 m) and PlanetScope for high-resolution (3 m) imagery, named after the optical images used for landslide delineation. Both subsets were designed based on a pool of 21 candidate features, including spectral bands, vegetation indices, SAR-derived backscatter metrics from Sentinel-1, and topographic attributes derived from the DEM published by the Geospatial Information Authority of Japan. A rigorous feature selection includes statistical and model-based evaluations, narrowing the list to the most significant features for landslide mapping, including green, red, NDVI, slope, and intensity.
Three deep learning models were employed on the JHRLD: UNet++, DeepLabv3+, and Medical Transformer (MedT). These models were evaluated using the F1 score for evaluating the JHRLD’s robustness and reliability. Performance analysis revealed that each model exhibited unique strengths depending on dataset resolution. On the moderate-resolution Sentinel-2 dataset, UNet++ excelled in detecting smaller-scale landslides, achieving an F1 score of 0.70. In contrast, DeepLabv3+ performed best on the high-resolution PlanetScope dataset, achieving an F1 score of 0.69 and effectively capturing large-scale and complex features. MedT showcased its superiority in boundary delineation, achieving the F1 score of 0.70 and excelling in identifying intricate landslide features. These results affirm the JHRLD’s robustness and reliability, providing a strong foundation for high-precision landslide detection across diverse resolutions and environments.
The JHRLD was validated in the Noto Peninsula, a region impacted by an Mw 7.5 earthquake and torrential rainfall in 2024. The model trained on the JHRLD demonstrated strong transferability, achieving an F1 score of 0.65 and a detection rate of 81% on this unseen area. Temporal and spatial analyses confirmed the JHRLD’s robustness, aligning well with observed hazard patterns under varying triggers.
The findings highlight the JHRLD’s adaptability as a benchmark dataset and its social utility in disaster prevention planning and emergency response.
Figure. Workflow for the JHRLD development and validation.
How to cite: Yu, B., Oguchi, T., and Iizuka, K.: Development and Validation of the Japan High-Resolution Landslide Dataset (JHRLD): Integrating Remote Sensing and Deep Learning, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9275, https://doi.org/10.5194/egusphere-egu25-9275, 2025.
The national database of mass movements includes more than 90000 landslide and snow avalanche events from the year 900 up to present. The main processes registered are landslides, snow avalanches, slushflows, ice falls and submarine landslides. The Norwegian Water Resources and Energy Directorate (NVE) has run the database since 2014 and can be accessed at www.skredregistrering.no and downloaded NVE - Nedlasting av kartdata
Some of the recorded landslides, like debris avalanches, debris flows, and shallow soil slides have been used to define landslide thresholds and create a landslide index that is used in the operational landslide forecasting and warning service. The poor quality of landslides registered in the database (uncertainty about landslide type, date and time and location) has limited the further analyses of thresholds and delayed the automatic updating of landslide thresholds. Controlled data could not be sent into the database; therefore, a new download of data and new control is necessary every time thresholds are to be updated. Could it be possible to give a quality score to each landslide, so it will be easy to automatically download data into threshold analyses, instead of a manual selection? Could it possible to perform the control directly into the database, to avoid a new control?
In recent years, procedures for quality control of historical landslides have been proposed at NVE, initially for weather-induced landslides, but later extended to other landslide types, like rock avalanches and clay slides, quick clay slides and slushflows. The quality control activity has been implemented more systematically since 2018 using all sources of information available: newspapers, aerial photos, technical reports. Four quality levels have been proposed, and quality criteria have been described. In this work it is presented how the quality control process is organized, which quality criteria are in use for the different landslide types, and which lessons we have learned.
The systematic quality control process is providing spatially and temporally improved dataset to be used not only in landslide threshold analyses, but also to be used in hazard mapping and the calibration of landslide models that simulate initiation areas and runout, used for susceptibility and hazard maps, as well as in other research projects. The control process has contributed to a better description and mapping of certain landslide types (like debris flows and debris avalanches, clay slides) and slushflows, and improved the overall registration and controlling tools.
How to cite: Devoli, G., Jensen, O. A., and Bekken, M. S.: Improving the quality of landslide events recorded in the Norwegian database , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10053, https://doi.org/10.5194/egusphere-egu25-10053, 2025.
Landslides remained the fatal disaster that contributed a number of human losses due to geological hazards, slope failures, debris flow and rockfalls in Malaysia. The cascading impact is further exacerbated by increasing magnitude and frequency of extreme weather events, climate change, and anthropogenic activities in urban settings, cultural heritage-affiliated, tourism-dominated areas, and food-insecurity regions in a tropical environment.
This study reports newly launched commitment and aspiration to reduce landslide disaster risk reduction and slope resilience strategies in Malaysia. The National Slope Master Plan (NSMP) 2025-2030 is an extended version of NSMP2009-2023, a 20-year road map with aims at enhancing the country’s capacity to assess and mitigate landslide risk, marking a significant milestone towards promoting sustainable slope management practices and reducing landslide disaster risk in Malaysia. The NSMP is collectively managed by the Department of Public Work under auspices of Inter-Governmental Agency Committee for Slope Management (ICSM). It is aligned to the international DRR agenda, UNDRR Sendai Framework for Disaster Risk Reduction 2015-2030 and Malaysia’s National Disaster Risk Reduction Policy 2030. Moreover, the NSMP Action Plan 2025-2030 incorporates a holistic framework, forward-looking and action-oriented guidance to integrate disaster risk reduction (DRR) into policies, programmes, development, and investments at all levels.
As a “living-document”, NSMP2030 serves a national guidance and primary reference for landslide disaster risk management at the national, local, and cross-sectoral levels. Remarkably, landslide and slope inventory remained a critical success factor to co-implement the multi-scale DRR plan. So far, we reported about 6441 landslides in the period of 1961-2024 and 25,608 slopes over mountainous environment, vulnerable highlands, tectonically active, hilly slope and urbanized settings. Two national guidelines are co-developed to address the multi-tier inventories for landslides and slope failures.
This study also explores new modalities of implementation, mean risk governance, inter-agency management mechanisms and integrated partnerships for de-risk investment. This study also highlights the use of nationally-supported and locally-led landslide inventories for supporting the development of highland vulnerability index (HVI) over 600,000 hectare in Cameron Highlands (Pahang), Kinta (Perak) and Lojing (Kelantan). It aims to enhance landslide disaster resilience in the vulnerable highlands by integrating comprehensive and inclusive localized DRR measures, promoting the well-being of the people, and supporting sustainable livelihoods and risk-informed development.
This transdisciplinary study emphasized progress made and achievements, to renew our shared commitments to amplifying and accelerating actions in all sectors and at all scale through 2030 and beyond, in pursuit of the global agenda and national aspiration in reducing disaster losses, preventing future systemic risk, and strengthening disaster resilience in a changing climate.
How to cite: Razak, K. A., Ramlee, L. H. S., Mei, S. Y., Razak, M. S. A., Hamidun, N., Ramli, Z., Mohamad, Z., Jaapar, R. A., and Ismail, M. F.: Landslide disaster risk reduction and slope resolience strategy in Malaysia, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-21464, https://doi.org/10.5194/egusphere-egu25-21464, 2025.
Landslide inventory maps (LIMs) are essential tools for hazard assessment, risk mitigation, and land-use planning. Expert knowledge significantly impacts their quality, potentially enhancing completeness and overall accuracy of landslide data. Experienced geomorphologists are trained to identify subtle topographic signatures of landslides, which is particularly the case of old or relict landslides or of complex geological settings, where the interpreters are supposed to deal with many sources of ambiguities.
This study examines the impact of expert knowledge on the quality of two geomorphological landslide inventory maps at a 1:10,000 scale in a geologically complex pilot area (45 km²) in Vinodol Valley, Croatia. The inventories were prepared through visual analysis of two LiDAR-based Digital Terrain Models (DTMs) at a resolution of 1m acquired in 2012 and in 2022. They were compared in terms of completeness, geographical accuracy, and thematic accuracy.
The first landslide inventory map (LIMA) was prepared by a single young researcher in 2018 using the 2012 LiDAR DTM. The second (LIMB) was prepared in 2024 using the 2022 LiDAR DTM by a team of three experts, including two geomorphologists with decadal experience in geomorphological mapping and the author of LIMA. Comparisons focused on the total number of landslides, completeness, degree of spatial agreement between the two maps, and landslide attributes, such as landslide classification, and relative age.
Results show that LIMA is incomplete compared to LIMB, especially when considering large and very large landslides, and LIMB includes more landslides, especially old and relict ones, which are mostly poorly visible on DTMs. We maintain that the incompleteness of LIMA, particularly focused on large, relict and less distinct landslides, can be attributed partially to the limited experience of the interpreter at the time of the mapping, and partially to the missing of a discussion approach in a multidisciplinary team.
This research highlights the importance of a collaborative approach in enhancing the quality of landslide inventory maps. While individual expertise is valuable, a diverse team of experts ensures more comprehensive and accurate mapping. Continuous training is essential to improve the detection of both recent and, especially, very old or relict landslides and to refine mapping skills necessary for accurate mapping in challenging environments.
How to cite: Jagodnik, P., Santangelo, M., Fiorucci, F., and Bernat Gazibara, S.: Expert knowledge in enhancing quality of landslide inventory maps: A LiDAR-based study from Vinodol Valley, Croatia, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9687, https://doi.org/10.5194/egusphere-egu25-9687, 2025.
Landslide inventory mapping over large, remote, and inaccessible areas remains a significant challenge due to the limitations of traditional field-based methods. Satellite-based InSAR (Interferometric Synthetic Aperture Radar) technology offers a viable solution by enabling the detection of surface displacements with millimeter-level precision, providing spatially extensive coverage of potentially landslide-prone areas. However, the accuracy of landslide detection using InSAR data can be compromised by the difficulty of distinguishing landslides from other types of surface deformation, such as subsidence, natural settlement, or deforestation, which can mimic landslide behavior in InSAR data. To address these challenges, we propose a machine learning-based approach that integrates InSAR-derived displacement time-series data with advanced pattern recognition techniques to identify and classify landslides, distinguishing them from ordinary ground movements. The methodology combines the high spatial and temporal resolution of InSAR with machine learning algorithms to recognize the distinctive features of landslides, such as sudden, non-linear displacements, velocity patterns, and deformation history. Feature engineering plays a crucial role, as key features like displacement rate, time-series patterns, and spatial characteristics (e.g., slope and curvature) are extracted from InSAR data to train machine learning models. These models can learn to differentiate between landslides and other ground movements by recognizing underlying patterns specific to landslide behavior. Supervised learning techniques are employed using labeled data (known landslide locations) to train models that can classify landslides accurately, even in areas with limited prior field data. In cases where labeled data is sparse, unsupervised learning techniques, such as clustering and anomaly detection, are applied to identify unusual displacement patterns that might indicate landslides. These models provide valuable insights into regions where landslides may occur, helping to distinguish between true landslide events and other non-landslide related surface changes. By integrating InSAR with machine learning-driven landslide pattern recognition, this approach enhances the accuracy and efficiency of landslide inventory mapping, particularly in large and remote areas where traditional field assessments are impractical. This methodology offers a scalable solution for early landslide detection, risk assessment, and hazard mapping. We discuss the potential benefits, challenges, and future directions of this approach, highlighting its applicability in diverse geographical settings and its role in advancing landslide monitoring and management strategies.
How to cite: Arsyad, A.: Machine Learning-Enhanced InSAR for Landslide Mapping: Differentiating Landslides from Ordinary Ground Movements, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3429, https://doi.org/10.5194/egusphere-egu25-3429, 2025.
Event landslide mapping plays a critical role in understanding the impact of triggering events, supporting emergency response, and defining risk reduction strategies. It also provides valuable datasets for validating susceptibility and risk models and training automatic landslide detection methods using machine learning. Enhancing our ability to detect and map landslides, particularly under challenging conditions, is key to improving response capacity during large-scale events.
Optical post-event images, while commonly used for mapping, are often unavailable immediately after a disaster due to dense cloud cover and limited revisit times. Synthetic Aperture Radar (SAR) sensors, with their ability to acquire data regardless of cloud cover or lighting conditions, offer a promising alternative.
In this study, we evaluated the reliability of L-band ALOS SAR amplitude images to prepare a landslide inventory map for the region affected by the MW 6.6 Hokkaido earthquake in September 2018, which triggered over 6,000 landslides. Using amplitude images of the radar backscattering coefficient (beta naught), we derived log-ratio change detection maps that highlight surface changes caused by landslides. These maps were visually interpreted by an expert geomorphologist to produce a SAR-based inventory in three test areas selected to represent varying landslide densities: low, medium, and high, as defined by the benchmark inventory.
To validate this approach, we compared the SAR-based inventory with a benchmark inventory derived from the interpretation of post-event optical images and field checks. The comparison assessed spatial coverage, geometric accuracy, completeness, and size distribution.
Results showed a good agreement between the SAR-based inventory and the benchmark, largely due to the high resolution of ALOS images, which enabled accurate detection and delineation of most landslide-affected areas. However, in the high-density test area, the delineation of individual landslides was less precise, with some generalizations observed. In contrast, the low-density test area exhibited more commission errors, likely due to challenges in distinguishing true landslides from noise in sparsely affected regions.
Our findings further demonstrate the potential of SAR data for landslide mapping in complex scenarios. The robust dataset produced in this study provides a rational basis for developing and training automatic landslide mapping systems based on radar backscattering log-ratio images.
How to cite: Santangelo, M., Mondini, A., Manconi, A., Iizuka, K., and Oguchi, T.: Event landslide mapping using L-band SAR data: insights from the 2018 Hokkaido earthquake, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15029, https://doi.org/10.5194/egusphere-egu25-15029, 2025.
How to cite: Furuki, H.: Research on the acquisition of topographic interpretation capabilities using deep learning and the generation of regional landslide susceptibility maps., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5550, https://doi.org/10.5194/egusphere-egu25-5550, 2025.
Due to the complexity of geological environments, hazards such as rockfalls, landslides, and debris flows often exhibit significant heterogeneity. Their spatial distributions typically display clustering across various scales. In this study, we propose a conditional probability-based model for assessing geological hazard susceptibility, which incorporates the cumulative effects of multiple geological environmental factors. This model is particularly suited for large regions with complex geological patterns.
To quantitatively evaluate the geological hazard susceptibility index for the study area, we first applied the Certainty Factor (CF) method to normalize 11 geological environmental factors within the same range. Subsequently, we introduced positive contribution (W⁺) and negative contribution (W⁻) parameters to measure the contribution of each factor to hazard occurrence. Using these parameters, we calculated the comprehensive influence coefficient (C) of each factor. The influence coefficients were then normalized to determine the weights of the geological environmental factors. Finally, the hazard susceptibility index (G) for the region was obtained by aggregating the CF values and their respective weights for the 11 factors.
Weights of Geological Environmental Factors for Hazard Susceptibility Assessment
| Geological Factors | Geological Environmental Factors | Weights |
| Topography and Geomorphology | Elevation（m） | 0.265 |
| Topography and Geomorphology | Slope（°） | 0.038 |
| Topography and Geomorphology | Aspect（°） | 0.026 |
| Lithology | Lithology Type | 0.010 |
| Geological Structure | Seismic Acceleration（g） | 0.090 |
| Geological Structure | Distance to Fault（m） | 0.003 |
| Meteorological and Hydrological Conditions | Hydrogeological Type | 0.036 |
| Meteorological and Hydrological Conditions | Water System Density（km/km2） | 0.093 |
| Meteorological and Hydrological Conditions | Annual Precipitation (mm) | 0.156 |
| Human Engineering Activities | Road Density（km/km2） | 0.119 |
| Human Engineering Activities |
Density of urban and large industrial buildings（one/km2） |
0.163 |
We applied this model to the Ili Valley region in Xinjiang, Northwest China, using data from 1,810 documented hazards. In the Geographic Information System (GIS) environment, we selected, processed, and analyzed 11 geological environmental factors, including elevation, slope angle, slope aspect, lithology, seismic acceleration, distance to faults, hydrogeological type, drainage density, annual rainfall, road density, and the density of urban and large civil infrastructure distributions. The model’s validation demonstrated reliable predictive performance for the study area. This research provides a practical method for evaluating geological hazard susceptibility, offering valuable insights for geohazard assessment and risk management.
Certainty Factor (CF) and Geological Hazard Susceptibility Index (G) Calculation Results for Geological Environmental Factors
How to cite: Yixi, W., Shouding, L., Shiwei, M., Xinshuo, C., Bo, Z., Tianqiao, M., and Xiao, L.: A Conditional Probability-Based Model for Geological Hazard Susceptibility Assessment, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5797, https://doi.org/10.5194/egusphere-egu25-5797, 2025.
The devastating earthquakes of 6 February 2023 in Türkiye (Mw 7.7 and Mw 7.6) triggered widespread co-seismic landslides across the region. This study focuses on developing and validating a landslide susceptibility map (LSM) for a 38,500 km² area in southeast Türkiye, which represents 5% of the country's landmass. Using a pre-earthquake inventory and the random forest algorithm, nine geomorphological and environmental features, including altitude, slope, lithology, and distance to faults, were integrated into the model. Validation was performed with a co-seismic landslide inventory comprising 2,611 landslides identified through pre- and post-earthquake aerial photogrammetric datasets.
Internal validation with the test data randomly split from the training dataset demonstrated high accuracy (93.67%) of the model based on the pixel-level assessments. However, the independent validation using co-seismic landslides revealed challenges, particularly in regions with rare lithological units or incomplete pre-event inventories. Despite the very limited pre-earthquake inventory, an accuracy of 76% was achieved, although it resulted in a significant number of false non-landslide labels. Thus, the co-seismic landslides highlighted the importance of accounting for unseen features, such as rare lithological units in the modeling. In addition, the resolution of the digital elevation model (EU-DEM with 25 m resolution) used for the LSM production was different from the resolution of the DEM used for post-earthquake landslide delineation. The latter one was obtained from high resolution aerial stereo images. Since the size of landslides which can be determined with the LSMs have strong correlation with the DEM quality, the difference between the internal accuracy and the external assessment results can be partly attributed to the data source used for inventory compilation. Nonetheless, the EU-DEM was found suitable for regional LSM production, and higher resolution DEMs also introduce computational complexity for such a large region.
This study outcomes revealed the potential of integrating remote sensing, machine learning, and geospatial data to enhance regional landslide susceptibility mapping. The findings provide valuable insights for disaster risk reduction, urban planning, and mitigating the impacts of latent hazards in seismically active regions; while pointing out the importance of data quality, optimization of machine learning algorithms, and multi-temporal inventory analyses to improve predictive accuracy.
How to cite: Kocaman, S., Karakas, G., Unal, E. O., Cetinkaya, S., Tunar Ozcan, N., Karakas, V. E., Can, R., and Gokceoglu, C.: Assessing Landslide Susceptibility Prediction Performance with an Event-Based Inventory from the 6 February 2023 Türkiye Earthquakes, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-20564, https://doi.org/10.5194/egusphere-egu25-20564, 2025.
Landslide catalogues have grown such that they allow for increasingly robust estimates of the size scaling of slope failures. Relationships between landslide volume, area, and their relative abundance provide useful insight into quantitative models of hillslope stability, hazard and risk, and landscape evolution. Numerous studies concur that smaller landslides are systematically more frequent than larger ones, and fitted various probability distributions to mapped landslide areas or volumes to capture this inverse relationship. However, especially the larger and commensurately rarer landslides (defined here as affecting footprint areas ≥0.1 km2) tend to eldude these statistical analyses. Thus, it remains unclear as to how an extrapolation of models derived from smaller landslides is valid beyond the size range identified for a given study area. Similarly, it can be problematic to use scaling statistics from other inventories because of likely differing methods of landslide detection and mapping, data quality, resolution, sample size, model choice, and fitting. We propose a multi-level Bayesian Generalised Pareto model as common ground for consistently estimating and comparing size distributions of large slope failures from different catalogues. The model remediates the problem of small sample size and makes use of all available data from thousands of landslides across several dozens of databases. The underlying peak-over-threshold approach is firmly rooted in extreme-value theory and offers a statistical reference against which any physical interpretations of landslide scaling statistics can be compared. We find that, despite a broad set of mapping protocols and lengths of record, and differing topographic, geological, and climatic conditions, the posterior power-law exponents remain indistinguishable between most inventories. The same goes for known earthquake from rainfall triggers, and event-based from multi-temporal catalogues. However, our model identifies several inventories with outlier scaling statistics that more likely result from censoring effects during the mapping or compilation process. We thus caution against a universal or solely mechanistic interpretation of scaling parameters, at least concerning large landslides. Some of this physical meaning might get diluted, mixed, or even lost in empirical data that combine confounding controls.
How to cite: Korup, O., Luna, L., and Joaquin, F.: Size and frequency of large landslides from different incomplete inventories, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15100, https://doi.org/10.5194/egusphere-egu25-15100, 2025.
Based on a minimum amount of rainfall that can trigger landslides when reached or exceeded, rainfall thresholds are used to predict the occurrence of rainfall-induced landslides and are an essential part of many landslide early warning systems worldwide.
The most common information used to define empirical rainfall thresholds is rainfall duration, cumulative rainfall and landslide occurrence time, all of which are derived from data sets with uncertainties, which are particularly important to consider when thresholds are used in early warning systems.
The landslide information is usually obtained from a variety of sources, including newspapers, blogs, landslide databases, scientific journals, technical documents, event and firefighter reports, and the association between the geographical location and time of occurrence of the landslides and the rainfall records is made by expert judgement based on heuristic criteria. Inaccuracies in the location and/or time of occurrence of the landslide and lack of systematic mapping are the main sources of uncertainty.
Assuming that a power law is a good descriptor of the dependence of cumulative rainfall on rainfall duration, in this work we focus our interest on a strategy to mitigate the epistemic uncertainties associated with the data that affect the model parameters: we propose an ensemble approach based on four different models to estimate the exceedance probability of landslide occurrence, which we combine through a voting scheme.
Methods include a frequentist ordinary least square regression method, a frequentist quantile regression method, a Bayesian quantile regression method, and a machine learning symbolic regression method.
The thresholds obtained by the four methods are equivalent to the opinions of four independent experts who were asked to give their advice on the minimum amount of cumulative rainfall required for a potential landslide to occur for a given duration of rainfall.
We measure the level of agreement among the experts by counting the number of predictions that are above, below or in the range of uncertainty of the four thresholds. Finally, we take the most voted prediction as representative of the rainfall condition and the level of agreement/disagreement as an indication of the uncertainty in our prediction.
This approach provides a novel and robust framework for considering uncertainty in rainfall thresholds and offers practical insights to enhance decision-making in landslide risk management.
How to cite: Melillo, M., Mondini, A., and Guzzetti, F.: Rainfall threshold ensemble for landslide prediction under data uncertainty, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-21121, https://doi.org/10.5194/egusphere-egu25-21121, 2025.
