Reliable loss assessment of tropical cyclones (TCs) is critical for effective disaster emergency response. Existing methods often overlook the combined impacts of multiple hazards associated with TCs, such as wind, rainfall, storm surge, waves, and floods, which can decrease loss estimation accuracy. To address this issue, a novel assessment framework is proposed that integrates these multi-hazard effects to enhance disaster loss modeling. This framework begins by identifying multi-hazard features of TCs, including maximum gust wind (3s), total rainfall, daily rainfall, hourly rainfall, surge heights, significant wave heights, and daily runoff. Using a dataset of 1,341 county-level records, four machine learning algorithms—Categorical Boosting (CatBoost), Transformer, Backpropagation Neural Network (BPNN), and Support Vector Machine (SVM)—are trained and optimized. The best-performing model is applied to assess the impact of feature variables and training samples. Additionally, shapley additive explanations (SHAP) are employed to interpret the model, providing insights into feature importance and relationships among hazards. Results indicate that CatBoost outperforms other algorithms, achieving an accuracy of 0.8196. Incorporating all feature variables results in a maximum performance improvement of 19.06% compared to using single, double, or triple hazards. The model demonstrates strong applicability across coastal and inland regions at the national scale, maintaining an accuracy above 0.79. By integrating SHAP analysis, this approach enhances model interpretability, offering valuable insights into factor contributions and inter-hazard relationships. The proposed framework improves the reliability of loss assessments and addresses the limitations of machine learning "black boxes," supporting more informed and effective disaster response strategies.

How to cite: Zheng, J., Fang, W., and Shao, J.: An interpretable multi-hazard machine learning model for county-level loss assessment of tropical cyclones, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-4761, https://doi.org/10.5194/egusphere-egu25-4761, 2025.

Despite the plethora of studies on landslide analysis and prediction, buildings are often the structures that endure the most tangible harm and must address the aftermath. In Korea, landslide damage attributable to climate change is escalating, particularly impacting buildings and residences. To mitigate this issue, it is imperative to forecast the areas where landslides are likely to occur and identify structures within their potential damage range. Consequently, this study aims to develop a landslide risk analysis model for buildings.

This landslide risk analysis model consists of three steps: (1) deriving landslide-susceptible areas, (2) deriving landslide damage areas, and (3) identifying buildings expected to be damaged by landslides.

To derive landslide-susceptible areas, data on past landslide occurrences and environmental variables related to topography, soil, vegetation, and climate were utilized. To enhance the reliability of the dependent variable, Pearson's correlation coefficient was employed to exclude variables with high intercorrelation. Machine-learning-based ensemble models—namely artificial neural networks (ANN), extreme gradient boosting (XGBoost), and generalized linear models (GLM)—were then applied to analyze these landslide-susceptible areas. The area under the curve (AUC) for the final model’s accuracy analysis was 0.934, indicating a high degree of predictive accuracy.

To derive the landslide damage area, various runout models were considered, and LAHARZ was ultimately selected as the analysis tool. LAHARZ, developed by the United States Geological Survey (USGS), can simulate debris flow behavior and is frequently used for landslide damage analysis. In this study, potential landslide initiation points—identified from the landslide-susceptible area results—were combined with weather, topography, geology, soil, and vegetation data to determine the extent of debris flow damage in the event of a landslide.

In the final stage of the analysis, buildings located within the debris-flow damage area were extracted. To achieve this, building register information was geocoded and converted into spatial data, using the geocoding tool on a selected sample area. The analysis revealed that in 10 of the 19 potential landslide sites, buildings are situated within the damage range in the event of a landslide. However, in the remaining 9 sites, no buildings are damaged even if a landslide occurs. Consequently, a total of 67 buildings in the sample area are likely to be damaged. These include 14 apartments, 6 multi-family/multi-unit houses, 2 single-family houses, and 1 apartment complex. The model developed in this study can serve as a foundation for residents and building users to respond more effectively to potential landslide damage.

How to cite: Song, Y. M., Cho, Y., and Kim, H. G.: Using Machine Learning and LAHARZ to Develop a Landslide Risk Analysis Model for Buildings, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5516, https://doi.org/10.5194/egusphere-egu25-5516, 2025.

Landslide susceptibility mapping is a vital tool for identifying areas vulnerable to slope instability and mitigating related hazards. A critical challenge in this process is constructing a robust, diverse, and balanced training dataset that accurately distinguishes landslide-prone areas from stable regions. This study proposes a methodology that integrates multi-buffer zoning, clustering-based sampling, and stratified sampling to enhance predictive accuracy and dataset representativeness.

The study was conducted in the Paphos district of Cyprus, an area of 552 km² that has experienced over 1,800 recorded landslides. The region’s geomorphological complexity, shaped by diverse topographic, geological, hydrological, and land-use conditions, makes it an ideal setting for advancing landslide susceptibility mapping techniques. A comprehensive dataset incorporating key environmental variables—such as slope, elevation, curvature, lithology, proximity to faults, and land cover—was compiled for analysis.

To develop the training dataset, documented landslide points were paired with non-landslide points generated from three spatial buffer zones: 250 m, 500 m, and 750 m around landslide sites. To further improve data diversity, clustering-based sampling grouped data points based on geomorphological and environmental similarities, while stratified sampling ensured proportional representation of critical variables in the dataset.

Three machine learning models—Logistic Regression (LR), Random Forest (RF), and XGBoost—were employed to evaluate the predictive performance of datasets constructed using individual buffer zones, clustering, and stratification techniques. Model performance was assessed using metrics such as Accuracy, F1 Score, Cohen’s Kappa, and Area Under the Curve (AUC) to determine the effectiveness of each dataset.

The results revealed clear distinctions between datasets. The 750 m buffer dataset outperformed the others, with XGBoost achieving an Accuracy of 93.92%, F1 Score of 93.86%, Cohen’s Kappa of 87.84%, and AUC of 98.36%. This dataset effectively captured stable environmental conditions, improving model robustness and generalizability. The 500 m buffer dataset also performed well, with XGBoost achieving an Accuracy of 92.36% and an AUC of 97.66%, while the 250 m buffer dataset, exhibited slightly lower performance, with XGBoost achieving an Accuracy of 89.36% and an AUC of 95.77%.

The clustering-based sampling approach also demonstrated strong results, with RF achieving an Accuracy of 92.44% and an AUC of 97.19%, suggesting that grouping data points based on shared characteristics enhances model precision. Finally, the combined dataset, which integrated clustering-based and stratified sampling, yielded robust results, with XGBoost achieving an Accuracy of 93.74%, Cohen’s Kappa of 85.99%, and AUC of 97.99%.

In conclusion, the proposed approach demonstrates the value of integrating multi-buffer zoning, clustering, and stratified sampling into susceptibility mapping frameworks. This study not only advances our understanding of landslide processes in the Paphos district but also provides a scalable, reliable methodology for landslide risk assessment in other regions, contributing to more resilient landscapes and communities.

This research was funded by the European Commission, project reference: ENTERPRISES/0223/Sub-Call1/0229

How to cite: Tsangaratos, P., Chrysafi, A.-A., Tzampoglou, P., Anastasiades, A., Valari, E., Giannoglou, V., and Loukidis, D.: Dataset Construction for Landslide Susceptibility Mapping Using Multi-Buffer Zones, Clustering, and Stratified Sampling, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13938, https://doi.org/10.5194/egusphere-egu25-13938, 2025.

Landslides rank among the most devastating natural hazards globally, causing widespread socio-economic disruptions and posing significant threats to human lives, infrastructure, and ecosystems. These events are primarily triggered by extreme weather conditions, such as heavy rainfall, and result from complex interactions between hydrological conditions, soil saturation, and terrain instability. This study focuses on southeastern Thessaly, specifically Mount Pelion in central Greece, a region with high geomorphological complexity and significant landslide susceptibility. Situated between the Aegean Sea and the Pagasetic Gulf, Mount Pelion’s diverse landscape, shaped by its unique climatic and geological features, makes it an ideal case study for exploring the relationships between morphometric and hydrological parameters and landslide activity.

The region's geological formations range from the Quaternary to the Triassic periods. While Quaternary deposits, composed mainly of sandy clays and gravels, are typically stable and found in torrent beds and coastal areas, the unstable Neo-Paleozoic to Triassic formations dominate the region. These formations, which include schists, quartzites, gneisses, and marbles, account for over 90% of historical landslides, highlighting their critical role in slope instability. 

This research presents a detailed geomorphological and hydrological analysis of 15 basins within the region, utilizing a variety of morphometric parameters. These include basin area, perimeter, elevation metrics, stream density, ruggedness indices, and shape indices like the Gravelius index and circularity ratio. Statistical analyses, including Pearson and Spearman correlation tests, were conducted to evaluate the influence of these parameters on landslide occurrences. The study also incorporated SHAP (SHapley Additive exPlanations) analysis to quantify the global impact of key features on landslide susceptibility predictions. 

Positive correlations between landslide occurrences and variables such as basin area (p: 0.981), stream length (p: 0.964), and perimeter (p: 0.948) emphasize the role of large basins with extensive hydrological networks and complex boundaries in increasing landslide susceptibility. Elevation metrics, including maximum elevation (p: 0.765) and mean elevation (p: 0.713), further underscore the vulnerability of high-altitude terrains with steep slopes. Conversely, negative correlations were observed for compact basin shapes (Gravelius index: p: -0.745, s: -0.923) and lower relief ratios (p: -0.676, s: -0.773), indicating that compact and less steep basins are less prone to landslides due to efficient runoff and reduced infiltration. The SHAP analysis further identified basin area (F), relief ratio (Rv), stream flow length (SF), and ruggedness index (Rn) as the most influential features driving landslide risk, with high values of these parameters significantly increasing susceptibility. Features like maximum elevation (Hmax) showed moderate positive impacts, while perimeter (P) and stream length (SL) exhibited lesser influence.

In conclusion, this study offers a robust framework for understanding the geomorphological behavior of basins and its impact on landslide susceptibility. By linking key parameters to slope instability, it contributes to the development of effective mitigation strategies and supports sustainable management of landslide-prone regions. Insights from this analysis hold practical value for disaster risk reduction, resource management, and long-term resilience planning in geologically complex landscapes like southeastern Thessaly.

How to cite: Chrysafi, A.-A., Ilia, I., Albano, R., Chen, W., Matiatos, I., and Tsangaratos, P.: Geomorphological and Hydrological Analysis of Landslide-Prone Basins: A Case Study from Mount Pelion, Central Greece, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-16033, https://doi.org/10.5194/egusphere-egu25-16033, 2025.

Groundwater overdraft in arid and semi-arid regions poses a significant threat to sustainable water resources. In Balochistan, Pakistan, a region with limited precipitation (<400 mm/yr) but high reliance on groundwater for agriculture and urban supply, excessive water extraction has led to dramatic land subsidence in the inhabited valleys. These deformations, have been documented since the 1990s. Using two-dimensional Interferometric Synthetic Aperture Radar (InSAR) analysis, we generated high-resolution surface deformation maps to characterize subsidence and its evolution over the Kharan drainage system between 2014 and 2024. Subsidence rates exceed 15 cm/year in urban centers like Quetta, while surrounding agricultural valleys show variable deformation patterns, including seasonal motion of about 2 cm. To identify dominant deformation modes, we applied independent component analysis (ICA) to decompose temporal signals, linking them to precipitation variability, groundwater level changes, and land use dynamics. Our results also highlight the potential role of faults in modulating aquifer connectivity and deformation patterns. By combining spatio-temporal deformation analyses with meteorological and geographic data, we provide insights into groundwater recharge, aquifer behavior, and the sustainability of water resources in the face of ongoing population growth and climate change.

How to cite: Dalaison, M., Chanard, K., Jolivet, R., Raimbault, B., and Kakar, N.: Decadal high-resolution mapping of land subsidence driven by severe groundwater overdraft in Balochistan, Pakistan, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-16369, https://doi.org/10.5194/egusphere-egu25-16369, 2025.

The availability of Earth Observation (EO) data, which are nowadays freely accessible to an increasing extent, has significantly advanced large-scale monitoring capabilities for geological hazards, particularly in terms of acquisition frequency and areal coverage. This progress has been especially evident in monitoring land subsidence. By the first quarter of 2022, the Copernicus European Ground Motion Service (EGMS) began providing ground displacement data at the European level, offering valuable insights into surface movements across the continent. Despite the growing use of Interferometric Synthetic Aperture Radar (InSAR) for monitoring land subsidence, relatively few studies have focused on translating this EO data into comprehensive risk assessments.

The goal of this work is to develop a novel EO-based methodology for mapping land subsidence risks at regional scale. This methodology has been tested in the Emilia-Romagna region of Italy, an area historically affected by land subsidence due to both natural processes and anthropogenic factors. In this region, land subsidence rates have reached up to 7 cm/year since the 1950s.

To estimate the exposure and vulnerability of the region, we have utilized data from the World Settlement Footprint (WSF) Evolution and the Global Human Settlement Layer (GHSL), both of which offer crucial insights into the human settlements and infrastructure that could be impacted by land subsidence. Moreover, we have exploited EGMS ground displacement data to estimate hazard levels associated with differential settlement. The resulting land subsidence risk map identifies four distinct risk levels, ranging from low to very high, across various areas of Emilia-Romagna. It offers a user-friendly product helping land use planners and local authorities to better understand and mitigate the potential impacts of land subsidence in the affected areas.

This work is funded by the European Union – Next Generation EU, component M4C2, in the framework of the Research Projects of Significant National Interest (PRIN) 2022 National Recovery and Resilience Plan (PNRR) Call, project SubRISK+ (www.subrisk.eu; grant id. P20222NW3E), 2023-2025 (CUP B53D23033400001).

How to cite: GoliRaeisi, L., Bonì, R., Taramelli, A., Cigna, F., Teatini, P., Paranuzio, R., and Zoccarato, C.: Integration of Earth Observation data into land subsidence risk mapping: the Emilia Romagna region case of study (Italy), EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-16710, https://doi.org/10.5194/egusphere-egu25-16710, 2025.

Forest anomalies (e.g., pests, deforestation, and fires) are common phenomena of the earth’s surface. Rapid detection of these anomalies is important for sustainable forest management and development. On-orbit remote sensing detection of multi-type forest anomalies using single-temporal images is one of the most promising methods for achieving it. Nevertheless, existing forest anomaly detection methods rely on time-series image analysis and are designed for a single type of forest anomaly. Here, a Forest Anomaly Comprehensive Index (FACI) was proposed to rapidly detect multi-type forest anomalies (i.e., pests, deforestation, and fires) using different thresholds and single-temporal Sentinel-2 images. First, the spectral characteristics of different forest anomaly events were analyzed to obtain potential band combinations for comprehensive anomalies detection. Then, the FACI form based on the potential bands was determined using images simulated by the LESS model. The threshold separability of FACI was compared to that of existing indices (NDVI, NDWI, SAVI, BSI, and TAI). In the evaluation, the thresholds for FACI and existing indices were determined using the interquartile method and 90 field survey samples, while their accuracy was quantitatively assessed with an additional 90 field survey samples and Sentinel-2 images. Finally, the evaluation results indicated that the overall accuracy of FACI in detecting the three forest anomalies was 88.3%, with the corresponding Kappa coefficient of 0.84. While all the overall accuracy of existing indices are below 80%, with Kappa coefficient less than 0.7. Meanwhile, a case study in Ji'an, Jiangxi Province confirmed the ability of FACI to detect different stages of pest infection, as well as the deforestation and forest fires using single-temporal satellite images. Overall, FACI represents a promising method for detecting multi-type forest anomalies in future real-time on-orbit satellite applications.

How to cite: Liang, D. and Cao, B.: A new remote sensing index for multi-type forest anomalies detection based on Sentinel-2 imagery, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17103, https://doi.org/10.5194/egusphere-egu25-17103, 2025.

Known as one of the world’s most dynamic deltas in terms of land-sea changes, the Yellow River Delta is rich in natural resources such as brine groundwater and oil. It is affected by tectonic movements, natural consolidation and compaction of loose sediments, and especially frequent anthropogenic activities. Consequently, various degrees of land subsidence occur, and the northern estuary region of the Yellow River Delta is one of the areas experiencing more intense land subsidence, presenting possible threats to the safety of local inhabitants and economic activities. Therefore, accurate monitoring and understanding the spatiotemporal distribution characteristics of land subsidence in the northern estuary region of the Yellow River Delta are of great significance to mitigate geological impacts and economic losses in the region. In this work, land subsidence information in the northern estuary region of the Yellow River Delta was obtained using InSAR time series technology, based on Sentinel-1A/B data collected from January 2020 to December 2021. Additionally, multi-source data, including soft soil thickness, precipitation, oil field and brine mining areas, were incorporated to identify the influencing factors and asses their relative importance in land subsidence through random forest analysis and post-interpretation techniques. The results show that land subsidence in the northern estuary region of the Yellow River Delta presents uneven distribution characteristics, exhibiting maximum annual average subsidence rate exceeding -100 mm/year. The results of the random forest model indicate that the primary factors influencing land subsidence in the northern estuary region of the Yellow River Delta are brine groundwater extraction and the thickness of the soft soil layer. Meanwhile, the post-interpretation analysis demonstrates changes in the relationships between the different influencing factors and land subsidence.

How to cite: Chen, M., Ge, P., Tomás, R., and Cheng, S.: Investigation of land subsidence in the northern estuary region of the Yellow River Delta, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18110, https://doi.org/10.5194/egusphere-egu25-18110, 2025.

The Urban Intelligence approach views the city as a complex system that needs to be studied through the interaction of its different subsystems. Such complexity is addressed also in the virtual dimension, through the construction of Urban Digital Twins that allow to understand, control, and optimize the urban dynamics according to multidimensional objectives.

In this context, we describe here a model to assess and evaluate the risks incurred by pedestrians and vehicles in a city under severe and extreme rainfall events that results in increasing of surface runoff, causing pluvial floods. This study is motivated by the increasing frequency of extreme events that seriously challenge the urban infrastructures in historical cities where urban design dates back centuries and constraints to structural modifications of the urban texture are often present.

The approach is based on the design and integration of two models: first, a traffic macro-simulation model that integrates multi-objective demand and resources in an optimal and automated way; such model, also referred to as the Mobility Digital Twin, can predict vehicle and pedestrian flows over the segments of the city network. Second, a model of water dynamics over the same city network (Water Digital Twin), based on the morphological structure of the territory and on the 3D urban model, that integrates a hydrological-hydraulic coupled model that is able, starting from predetermined rainfall events, to estimate the water levels and flow rates in each portion of the investigated territory of rainfall.

The two models are jointly used to create scenarios for different weather conditions, simulate recovery policies, identify the system’s bottlenecks and design evacuation strategies, both at the strategic and at the operational level. The results of the experimentation will be analyzed and implemented within the SIT. Specifically, with Intelligent SIT, we define a framework for integrating data from diverse sources, including informative, participatory, and human-centric data, as well as outputs from Thematic Digital Twins and other sources. To accurately represent complex systems, we rely on detailed maps and in-depth spatial analysis, made possible through the capabilities of the SIT.

A prototype application of the approach is developed for the City of Matera, within the Casa delle Tecnologie Emergenti project and the development of the city’s Urban Digital Twin. Preliminary results validate the potential contribution of the models adopted and have been used to support local authorities in the design of recovery strategies in the presence of extreme weather events and in the planning of  mitigation actions on the city road network.

Acknowledgments This research was supported by the “Casa delle Tecnologie Emergenti di Matera” project.

How to cite: Presta, I. G., Felici, G., Vitale, M., Stecca, G., Gaibisso, C., Martino, B. L., Albano, R., Ermini, R., and Castelli, G.: Integration of two models, Mobility Digital Twin and Water Digital Twin – Matera Case Study, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19077, https://doi.org/10.5194/egusphere-egu25-19077, 2025.

The quasi-cyclic behavior of fault zones, which encompasses interseismic, coseismic, and postseismic phases, is essential for understanding the dynamic evolution of earthquakes. Examining the spatiotemporal evolution of surface deformation and fault slip distribution across different periods of an earthquake cycle, obtained through geodetic techniques, enables a systematic and precise understanding of earthquake deformation models. The 2020 Elaziğ earthquake and the 2023 Kahramanmaraş Earthquake Sequence, both of which occurred along the East Anatolian Fault (EAF) in eastern Anatolia, provide a unique opportunity for studying the earthquake cycle and associated fault behavior. Historically, seismic activity along the EAF has demonstrated the fault’s capacity to produce significant earthquakes, with distinctive fault mechanisms varying across time and fault segments. In addition, approximately a decade of high-resolution surface deformation data obtained from InSAR, spanning five distinct periods in the earthquake cycle, is available for in-depth analysis.

Our objective was to provide a comprehensive characterization of present-day kinematic processes along the EAF, to gain insights into fault frictional properties and to assess potential future seismic hazards. To do so, we utilized high-resolution interferometric data to investigate fault slip evolution from March 2015 to June 2024. This temporally continuous deformation field allowed us to explore fault behavior and develop complete slip distribution models throughout the earthquake cycle, which includes an interseismic period (2015-2020), two postseismic periods (2020-2023 and 2023-2024), and three coseismic events. Initially, we conducted an InSAR time series analysis to capture the deformation fields across different periods of the EAF earthquake cycle. We then integrated high-resolution ground displacement data, aftershock distributions from the 2020 and 2023 earthquakes, and the Global Active Faults Database (GEM) to map the complex fault geometries of the EAF. These inputs facilitated the creation of triangular dislocation models for analyzing fault slip distribution at various periods of the earthquake cycle. Moreover, we examined the relationship between slip distribution and estimated frictional parameters along the EAF, followed by an assessment of seismic hazard potential.

Our analysis of slip evolution reveals that the postseismic fault slip following the 2020 and 2023 earthquakes primarily occurred in areas with minimal coseismic slip. We also identified four slip deficit regions, comprising both shallow and deep portions of the seismogenic faults. By integrating slip distributions and historical earthquakes, we calculated the total moment deficit rate for each fault segment, revealing that the Palu segment, as well as the central portions of the Erkenek and Sürgü-Çardak segment, possesses a high earthquake potential. These findings underscore the critical need for high-resolution and continuous monitoring of fault systems across different seismic periods, offering new insights into the dynamics of the earthquake cycle along the EAF.

How to cite: Han, B., Song, C., and Aoki, Y.: Fault spatial heterogeneity and seismic hazards revealed by geodetic observations of the East Anatolian Fault, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-95, https://doi.org/10.5194/egusphere-egu25-95, 2025.

Large-scale key projects such as pumped storage, wind farm or infrastructure projects are gradually increasing in the higher altitudes bare rock mountainous areas in China due to the national strategy. The high-level dangerous rock mass widely distributed in these areas poses a great threat to engineering construction due to its huge-scale and concealment. However, the traditional geological survey method is difficult to obtain the complete feature information of dangerous rock mass efficiently and accurately. Therefore, we optimized the data acquisition parameters of the oblique photography of dangerous rock mass in this special geological environment, and formed a set of targeted data acquisition ideas. Then, based on the oblique photography model, the interpretation signs of dangerous rock mass are established, and a set of identification and classification theory is summarized. Using point cloud data, the automatic identification technology of dangerous rock mass structural plane is further studied. At the same time, the research also carried out the application practice based on the actual pumped storage project, and verified the effectiveness and accuracy of the proposed method.This study showed that oblique photography is a promising method for improving high-level dangerous rock mass identification efficiency in bare rock mountainous areas.

How to cite: He, M., Peng, J., Ma, P., and Jia, Z.: Enhancing Dangerous Rock Mass Identification in Bare Rock Mountainous Areas Using Oblique Photography, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-4795, https://doi.org/10.5194/egusphere-egu25-4795, 2025.