Landslide Susceptibility Research Articles

Landslide susceptibility evaluation based on conditioning factor system construction and machine learning model optimization

ABSTRACT Machine learning models have been widely used in landslide susceptibility research, but issues such as unreasonable division of conditioning factor intervals, uneven quality levels of input datasets, and overfitting of machine learning models remain hot and difficult problems in the landslide susceptibility evaluation process. Aiming at the problems existing in the current research, this article takes Fengjie County, Three Gorges Reservoir Area, a high incidence area of landslide hazards, as the study area. Based on the geographic detector model, an interval division method suitable for a single conditioning factor in the research area is selected. The beetle algorithm is added to three machine learning methods to adjust hyperparameters and further optimize the model, completing the landslide susceptibility evaluation of the study area. By analyzing the accuracy, precision, recall, F1-Score and ROC curve, three models all show good fitting ability and evaluation ability. The AUC value of the DBO-XGBoost model (0.9628) is the highest, followed by the DBO-SVM model (0.9596) and the DBO-RF model (0.9457), which can provide an effective reference for hazard prevention and control in Fengjie County.

GIS-based landslide susceptibility mapping using Frequency ratio method: A case study from Adigrat-Mugulat mountain chains, northern Ethiopia

GIS-based landslide susceptibility mapping using Frequency ratio method: A case study from Adigrat-Mugulat mountain chains, northern Ethiopia

Analisis Kerawanan Longsor Berbasis Pemetaan Geomorfologi di Manggarai Timur

Landslide Susceptibility Analysis Using Geomorphological Mapping Approach in East Manggarai Region. Technology (IT), particularly Geographic Information Systems (GIS), plays a significant role in supporting the management of natural disasters such as landslides. East Manggarai has geographical conditions that are highly vulnerable to landslides, with topography dominated by steep hills and sharp slopes. To address this issue, it is necessary to analyze landslide susceptibility levels in the East Manggarai region using a geomorphological mapping approach through Geographic Information Systems (GIS). The mapping provides important contributions to the government in spatial planning, disaster prevention, and landslide risk mitigation. Landslide susceptibility maps and data will be published through an interactive website, facilitating information access for communities and stakeholders. This is expected to help communities identify areas prone to landslides and predict potential landslide occurrences.

Examination of Landslide Susceptibility Modeling Using Ensemble Learning and Factor Engineering

Current research lacks an in-depth exploration of ensemble learning and factor engineering applications in regard to landslide susceptibility modeling. In the Three Gorges Reservoir area of China, a region prone to frequent landslides that endanger lives and infrastructure, this study advances landslide susceptibility prediction by integrating ensemble learning with systematic factor engineering. Four homogeneous ensemble models (random forest, XGBoost, LightGBM, and CatBoost) and two heterogeneous ensembles (bagging and stacking) were implemented to evaluate 14 influencing factors. The key results demonstrate the Digital Elevation Model (DEM) as the dominant factor, while the stacking ensemble achieved superior performance (AUC = 0.876), outperforming single models by 4.4%. Iterative factor elimination and hyperparameter tuning increased the high-susceptibility zones in the stacking predictions to 42.54% and enhanced XGBoost’s low-susceptibility classification accuracy from 12.96% to 13.57%. The optimized models were used to generate a high-resolution landslide susceptibility map, identifying 23.8% of the northern and central regions as high-susceptibility areas, compared to only 9.3% as eastern and southern low-susceptibility zones. This methodology improved the prediction accuracy by 12–18% in comparison to a single model, providing actionable insights for landslide risk mitigation.

Impact of plasticity and stress history on thermal volume changes in clays

This endeavor explores fine-grained soils’ thermally induced volumetric behavior through a series of temperature-controlled oedometer experiments under drained conditions. Undisturbed clay samples were subjected to incremental heating and cooling to evaluate the effects of over-consolidation ratio (OCR), stress history, and soil plasticity. Results revealed that normally consolidated clays undergo significant plastic contraction during heating. Over-consolidated samples showed contraction-dominated responses, highlighting the limitations of OCR as a standalone predictor of thermal behavior, with stress history emerging as a key factor. Furthermore, the influence of soil plasticity was pronounced, with high-plasticity clays experiencing greater thermal contraction due to enhanced microstructural rearrangement and mineralogical effects. The heating and cooling cycle further highlighted the irreversibility of volumetric changes in normally consolidated states, while over-consolidated samples exhibited reduced thermal hysteresis. These findings offer a detailed understanding of thermally induced volume changes in fine-grained soils, revealing the interplay between stress state, consolidation history, and intrinsic soil properties. The insights gained are foundational for advancing predictive models, optimizing the design of thermally loaded geo-energy systems, and addressing climate-driven challenges such as soil-atmosphere interactions and landslide susceptibility.

Validating and enhancing data-driven landslide susceptibility prediction by model explanation and MT-InSAR techniques

ABSTRACT Data-driven landslide susceptibility assessment (LSA) remains unconvincing owing to the disconnection from the modelling to physical cognition of landslide causation. Most models are mere good fits to certain datasets and can produce unexpected bias in their prediction, misleading high-risk area zoning. To validate data-driven LSA, this study delved into the innate interactions between input landslide features and predictions by model explanation and compared feature permutation results with landslide statistical priors. Furthermore, multi-temporal interferometric synthetic aperture radar (MT-InSAR) derived ground deformation was applied in an alpha pixel fusion and growth method for LSA enhancement. This study took Hong Kong as the research area, employed extreme gradient boosting (XGBoost) for LSA, and utilized Shapley additive explanations (SHAP) method for model explanation. The mean Shapley values – indicating feature importance – for slope, stream power index (SPI) and land use are 1.31, 1.12, and 0.67, respectively. This aligns with the landslide feature permutation derived from prior statistics, verifying the model prediction reliability. The applied InSAR enhancement results in a 13% increase of ‘(very) high’ landslide susceptibility areas. Additionally, LSA results and ground deformation map were cross-validated in the virtual geographic environment. This study improves the reliability of data-driven LSA through model explanation and InSAR enhancement.

A comprehensive analysis of landslide susceptibility in Iyidere Basin (NE, Turkey) using machine learning techniques and statistical bivariate methods

Abstract Natural events are called disasters when they cause great damage, human suffering, or loss of life. Landslides, one of these disasters, cause significant damage to property and infrastructure and pose risks to people's lives. In this research, landslide susceptibility was studied in Iyidere Basin, located in northeastern Turkey. This basin, which is among the cities where the most landslide events occur in Turkey, is a very important representative area in terms of a comprehensive analysis of landslides in the region. Bivariate (frequency ratio, weight of evidence, statistical index) and machine learning methods (artificial neural network, logistic regression) were used to evaluate landslide susceptibility with fifteen environmental parameters and 588 landslide inventory data. Landslide inventory data was generated using different sources, and environmental parameters databases were created using various sources and software. A receiver operating characteristic curve and Kappa statistic value were generated to test the performance and reliability of the susceptibility maps. It was determined that landslide susceptibility is higher in the downstream part of the basin. Although it varies between methods, it has been determined that approximately one-quarter of the basin has high and very high landslide susceptibility. The most effective parameters (drainage density, slope, curvature, lithology, land cover, distance to stream, and roads) for susceptibility and their classes were revealed.

Comparative landslide susceptibility mapping using local inventories: a case study from Trabzon, Türkiye

Abstract This paper presents a novel approach to landslide susceptibility mapping by integrating two landslide inventories prepared by different national agencies of Türkiye (MTA and AFAD) in the Trabzon region. By leveraging these different inventories, the study aims to offer a more comprehensive understanding of landslide risks, addressing limitations in previous susceptibility models that typically rely on single inventory sources. Employing the frequency ratio method, the paper generates susceptibility maps from each database to examine the influence of landslides across various environmental factors. Additionally, an Analytical Hierarchy Process (AHP)-based map, incorporating environmental characteristics, literature, and expert opinions, is developed to provide a third perspective, independent of historical landslide data. The results indicate that AHP model classifies approximately 19.20% of the study area as very high and high susceptibility. In contrast, the MTA and AFAD models assign only 12.40% and 8.80% to high and very high categories, with most areas falling into low to moderate susceptibility. Comparisons with the Global Landslide Hazard Map further highlight the strengths and limitations of localized versus global assessments. This study contributes to the field by demonstrating the benefits of a dual-inventory approach, enhancing the precision of landslide susceptibility maps and providing valuable insights for disaster risk management and sustainable land-use planning.

Landslide susceptibility mapping for western coastal districts of India using geospatial techniques and eXplainable artificial intelligence

Landslide susceptibility mapping for western coastal districts of India using geospatial techniques and eXplainable artificial intelligence

Landslide spatial prediction using data-driven based statistical and hybrid computational intelligence algorithms

Optimization of landslide susceptibility model driven by geological environment: a key challenge for disaster reduction in mountainous areas. Xiaojin County in China has complex geology and active hazards, posing a threat to human and economic security. This study evaluated landslide susceptibility in Xiaojin County with high terrain heterogeneity by comparing three data-driven models - WoE, optimized RF, and RBFNetworks. The spatial correlation between 12 conditioning factors and landslides was analyzed, and the multicollinearity between factors was determined, and the significance of the factors was quantified by MDA and MDG, especially for elevation, soil, and distance to roads. The WoE model exhibits exceptional performance (AUC: 0.899 for training, 0.892 for validation), outperforming RF (AUC: 0.880 for training, 0.874 for validation) and RBFNetwork (AUC: 0.866 for training, 0.863 for validation). The results have significant implications for land development and management in Xiaojin County, while also challenging the machine learning paradigm.

Investigating the role of rock weathering and clay mineralogy in landslide occurrences within the exposed Tertiary Formations in Rangamati area

BackgroundLandslides are a major and recurring hazard in the folded Tertiary sedimentary rock units of Southeastern Bangladesh. The mechanical strength and stability of these rocks are strongly influenced by their mineralogical composition and geochemical alteration, which can affect their susceptibility to weathering and collapse. However, the role of these factors in triggering landslides remains poorly understood.ObjectiveThis research addresses this gap by linking rock weathering intensity and clay mineral composition to landslide mechanisms in the outcropped Tertiary Formations in the Rangamati region.MethodsDetailed geological field work has been conducted on four major exposed Tertiary Formations: Dupi Tila Formation, Tipam Sandstone Formation, Boka Bil Formation, and Bhuban Formation. A comprehensive analysis of 30 landslides is carried out to evaluate the geological influence, with 22 samples (both fresh and weathered) examined using X-ray Diffraction (XRD), X-ray Fluorescence (XRF), and Scanning Electron Microscope (SEM) techniques to determine clay content and the extent of weathering. Shale percentages are estimated from outcrops as well as from wireline logs using empirical equations.ResultsThe Bhuban Formation exhibits the highest shale content, followed by the Boka bil, Dupi Tila, and Tipam Sandstone formations. Illite is identified as the dominant clay type and they are more prevalent in the older formations. Smectite is also present in varying proportions contributing significantly to landslide occurrences through its expansive properties. The Upper Marine Shale in Bokabil Formation contains the most smectite, followed by the Dupi Tila and Bhuban formations. The degree of weathering is evaluated through field observations and oxide analysis, with average Chemical Index of Alteration (CIA) values exceeding 75, indicating intense weathering in all formations, as clearly reflected in the outcrops. Four primary types of landslides: flow, fall, slide, and complex—are identified across all formations in varying proportions. Flow is predominant in formations dominated by single rock type, such as the Bhuban and Tipam Sandstone formations, while slides are more common in formations with alternating layers of sandstone and shale, such as the Boka bil and Dupi Tila formations.ConclusionThe findings highlight that higher shale volume, clay content, and CIA values significantly elevate the landslide susceptibility of geological formations, especially when these factors are compounded by primary triggers like intense rainfall and human-induced slope modifications.

Population and Landslide Risk Evolution in Long Time Series: Case Study of the Valencian Community (1920–2021)

Assessing the size and situation of the population exposed to natural hazards is a fundamental step in addressing natural hazard management and emergency planning. Although much progress has been made in recent years in population geolocation by competent public bodies, gathering historical data beyond the present century to learn about the sequential evolution the affected population has experienced remains a difficult task. The recent publication of a historical population grid with adequate resolution allows progress to be made in resolving this problem. This paper is based on these data together with a map of landslide susceptibility in the study area and on the abundant resources provided by the Spanish Cadastre on dates of construction, surface area, and location of built plots. The size of the residential area built in the risk zone and its affected population was calculated since the early 1900s and with a decennial sequence. The risk to the population has been found to be stable or decreasing slightly over the entire historical series in the study area. However, the intensive tourism in some coastal municipalities in the north of Alicante has led to the construction of holiday homes in unsuitable locations in mountainous areas and with it an appreciable increase in risk.

Demystifying the predictive capability of advanced heterogeneous machine learning ensembles for landslide susceptibility assessment and mapping in the Eastern Himalayan Region, India

Demystifying the predictive capability of advanced heterogeneous machine learning ensembles for landslide susceptibility assessment and mapping in the Eastern Himalayan Region, India

Multi‐Criteria Decision Analysis Framework for Landslide Susceptibility Mapping With Analytical Hierarchy Process in Parts of Assam–Arakan Fold Belt, India

ABSTRACTLandslide is one of the most sought‐after research areas in the current multidisciplinary geoscience studies. In this study, a landslide susceptibility map was developed using a Multi‐Criteria Decision Analysis (MCDM) framework with the Analytical Hierarchy Process (AHP) in the Dima Hasao district of Assam, India. The region is located in the western extent of the Assam–Arakan fold belt. It is characterised by extensively folded and faulted mountainous terrain. The region is highly susceptible to landslide hazard due to several factors including climatic factors, weak lithologic conditions, anthropogenic activity and so forth. This study utilised a GIS‐based approach of mapping landslide susceptibility zones; geoprocessing of 12 landslide causative factors was considered, namely slope, rainfall, elevation, geology, distance from fault, drainage density, distance from drainage, curvature (Cu), distance from road, Land Use/Land Cover, Topographic Wetness Index and aspect were utilised by the AHP algorithm. Weightage criteria are assigned to all the 12 factors on a scale of 1–5 based on the significance of landslide occurrence. The AHP algorithm classified four landslide susceptible zones of different hazard, that is, low, moderate, high and very high susceptible zones. The obtained landslide susceptibility model was validated using the area under the curve (AUC) of the receiver operating characteristics (ROC) with an accuracy of 0.864, which confirmed its reliability.

Landslide susceptibility mapping using an entropy index-based negative sample selection strategy: A case study of Luolong county.

Landslides constitute a significant geological hazard in China, particularly in high-altitude regions like the Himalayas, where the challenging environmental conditions impede field surveys. This research utilizes the IOE model to refine non-landslide samples and integrates it with multiple machine learning models to conduct a comprehensive assessment of landslide susceptibility in Luolong County, Tibet. The IOE model objectively assigns weights to conditioning factors based on the degree of data dispersion, thereby enhancing the predictive accuracy when combined with machine learning models. This research employed Google Earth satellite imagery to construct a comprehensive database comprising 2517 landslide debris in Luolong County. Twelve conditioning factors were identified, encompassing geological environment, topography, meteorology, hydrology, vegetation, soil, and human activities. The IOE model was integrated with SVC, MLP, LDA, and LR models to systematically evaluate landslide susceptibility in Luolong County. The results demonstrate that, after optimizing the non-landslide samples, the coupled models significantly outperformed the unoptimized models in terms of AUC, accuracy, precision, and F1 score. The ranking of classification performance and effect among the four coupled models is IOE-MLP > IOE-SVC > IOE-LR > IOE-LDA. Notably, the AUC value of the IOE-MLP coupled model increased from 0.8172 to 0.9747. Moreover, in the extremely high susceptibility zones, the IOE-MLP model had the highest landslide frequency ratio among the four coupled models, demonstrating the optimal classification performance and the best classification effect. The study identifies land use, elevation, and slope as the predominant controlling factors conditioning landslides in Luolong County. The regions with the highest susceptibility to landslides in Luolong County are predominantly situated in the central areas near rivers and roads, whereas the areas with the lowest susceptibility are largely located in the southwestern, northern, and certain central regions at elevations above 4500 m, which are consistently shrouded in snow and ice. This comprehensive method effectively resolves the challenge of selecting non-landslide samples, thereby improving the predictive accuracy of the landslide susceptibility model. The results of this study offer significant insights for disaster prevention, mitigation, and land use planning in analogous geological settings.

Assessing Flood and Landslide Susceptibility Using XGBoost: Case Study of the Basento River in Southern Italy

Floods and landslides are two distinct natural phenomena influenced by different conditioning factors, though some environmental triggers may overlap. This study applied eXtreme Gradient Boosting (XGBoost) to develop susceptibility maps for both phenomena, using a unified approach based on the same geospatial predictors. The approach integrated topographical, geological, and remote sensing datasets. Flood event data were collected from institutional sources using multi-source and high-resolution remotely sensed data. The landslide inventory was compiled based on historical records and geomorphological analysis. Key conditioning factors such as elevation, slope, lithology, and land cover were analyzed to identify areas prone to floods and landslides. The methodology was applied to the Basento River basin in Southern Italy, a region frequently impacted by both hazards, to assess its vulnerability and inform risk management strategies. While flood susceptibility is primarily associated with low-lying areas near river networks, landslides are more influenced by steep slopes and geological instability. The XGBoost model achieved a classification accuracy close to 1 for flood-prone areas and 0.92 for landslide-prone areas. Results showed that flood susceptibility was primarily associated with low Elevation and Relative Elevation, and high Drainage Density, whereas landslide susceptibility was more influenced by a broader and balanced set of factors, including Elevation, Drainage Density, Relative Elevation, Distance and Lithology. The resulting susceptibility maps offered critical approaches for land use planning, emergency management, and risk mitigation. Overall, the results demonstrated the effectiveness of XGBoost in multi-hazard assessments, offering a scalable and transferable approach for similar at-risk regions worldwide.

Utilizing Remote Sensing and Random Forests to Identify Optimal Land Use Scenarios and Address the Increase in Landslide Susceptibility

Massive land use changes in Indonesia driven by deforestation, agricultural expansion, and urbanization have significantly increased landslide susceptibility in upper watersheds. This study focuses on the Sumber Brantas and Kali Konto sub-watersheds where rapid land conversion has destabilized slopes and disrupted ecological balance. By integrating remote sensing, Cellular Automata-Markov (CA-Markov), and Random Forest (RF) models, the research aims to identify optimal land use scenarios for mitigating landslide hazards. Three scenarios were analyzed: business as usual (BAU), land capability classification (LCC), and regional spatial planning (RSP) using 400 field-validated landslide data points alongside 22 topographic, geological, environmental, and anthropogenic parameters. Land use analysis from 2017 to 2022 revealed a 1% decline in natural forest cover, which corresponded to a 1% increase in high and very high landslide hazard areas. From 2017 to 2022, landslide risk increased as the “High” category rose from 33.95% to 37.59% and “Very High” from 10.24% to 12.18%; under BAU 2025, they reached 40.89% and 12.48%, while RSP and LCC reduced the “High” category to 44.12% and 34.44%, respectively. These findings highlight the critical role of integrating geospatial analysis and machine learning in regional planning to promote sustainable land use, reduce landslide hazards, and enhance watershed resilience with high model accuracy (>81%).

An integrated landslide susceptibility assessment in the Karakoram Mountains based on SBAS-InSAR and machine learning: a case study of the Hunza Valley

An integrated landslide susceptibility assessment in the Karakoram Mountains based on SBAS-InSAR and machine learning: a case study of the Hunza Valley

Effects of different division methods of landslide susceptibility levels on regional landslide susceptibility mapping

Effects of different division methods of landslide susceptibility levels on regional landslide susceptibility mapping

Factors influencing landslide occurrence in low-relief formerly glaciated landscapes: landslide inventory and susceptibility analysis in Minnesota, USA

Abstract In landscapes recently impacted by continental glaciation, landslides may occur where topographic relief has been generated by the drainage of glacial lakes and ensuing post-glacial fluvial network development into unconsolidated glacially derived sediments and exhumed bedrock. To investigate linkages among environmental variables, post-glacial landscape development, and landslides, we created a landslide inventory of nearly 10,000 landslides in five regions of the formerly glaciated low-relief state of Minnesota, United States. Multivariate logistic regression indicates the importance of slope angle, lithology, and the development of stream valleys to landslide distribution. Areas underlain by fine-grained glaciolacustrine and nearshore deposits that are incised by streams are particularly prone to shallow (< 1–2 m depth) landslides. Landslides also occur in a wide range of glacial and fluvial deposits, and as rockfall in layered Paleozoic sedimentary rocks in central and southern Minnesota and Precambrian igneous and sedimentary rocks in northeastern Minnesota. Although no more than 1–2% of the studied regions are susceptible to landslides, they can pose risk to life and safety, damage infrastructure, and impact water quality. The combination of recently generated low-relief steep slopes, extensive unconsolidated sediments, and layered sedimentary bedrock make this formerly glaciated landscape more susceptible to landslides than current national-scale models indicate.