Landslide Data Research Articles

A New Integrated Approach for Landslide Data Balancing and Spatial Prediction Based on Generative Adversarial Networks (GAN)

Landslide susceptibility mapping has significantly progressed with improvements in machine learning techniques. However, the inventory/data imbalance (DI) problem remains one of the challenges in this domain. This problem exists as a good quality landslide inventory map, including a complete record of historical data, is difficult or expensive to collect. As such, this can considerably affect one’s ability to obtain a sufficient inventory or representative samples. This research developed a new approach based on generative adversarial networks (GAN) to correct imbalanced landslide datasets. The proposed method was tested at Chukha Dzongkhag, Bhutan, one of the most frequent landslide prone areas in the Himalayan region. The proposed approach was then compared with the standard methods such as the synthetic minority oversampling technique (SMOTE), dense imbalanced sampling, and sparse sampling (i.e., producing non-landslide samples as many as landslide samples). The comparisons were based on five machine learning models, including artificial neural networks (ANN), random forests (RF), decision trees (DT), k-nearest neighbours (kNN), and the support vector machine (SVM). The model evaluation was carried out based on overall accuracy (OA), Kappa Index, F1-score, and area under receiver operating characteristic curves (AUROC). The spatial database was established with a total of 269 landslides and 10 conditioning factors, including altitude, slope, aspect, total curvature, slope length, lithology, distance from the road, distance from the stream, topographic wetness index (TWI), and sediment transport index (STI). The findings of this study have shown that both GAN and SMOTE data balancing approaches have helped to improve the accuracy of machine learning models. According to AUROC, the GAN method was able to boost the models by reaching the maximum accuracy of ANN (0.918), RF (0.933), DT (0.927), kNN (0.878), and SVM (0.907) when default parameters used. With the optimum parameters, all models performed best with GAN at their highest accuracy of ANN (0.927), RF (0.943), DT (0.923) and kNN (0.889), except SVM obtained the highest accuracy of (0.906) with SMOTE. Our finding suggests that RF balanced with GAN can provide the most reasonable criterion for landslide prediction. This research indicates that landslide data balancing may substantially affect the predictive capabilities of machine learning models. Therefore, the issue of DI in the spatial prediction of landslides should not be ignored. Future studies could explore other generative models for landslide data balancing. By using state-of-the-art GAN, the proposed model can be considered in the areas where the data are limited or imbalanced.

Landslide Susceptibility Zonation of Rongga District and Surrounding Areas Using Weight of Evidence (WoE) Method

Rongga District is located on West Bandung Regency, West Java, which is prone to landslide disaster. Morphological conditions in the form of steep hills become the one of landslide controlling factors. There are many landslide occurrences happen in this area, such as Nyomplong, Cibitung Village on March 23, 2020. The incident was triggered by heavy rain and strong winds. This area was chosen to assess the landslide susceptibility using the Weight of Evidence (WoE) Method. WoE is probabilistic bivariate method which connecting parameters causes landslide against distribution of landslide in research area. Landslide data which generated from direct observation in the field and satellite imagery morphology are 572 landslide events. The data is divided into two groups, the analysis data set (70%) and the validation data set (30%). The parameters used in the analysis are land use, slope, slope direction, curvature, elevation, rainfall, lithology, NDWI, NDVI, distance from road, distance from the river, distance to lineament, flow direction, lineament density, stream density and river density. The parameters validated by determining the value of the area under curve (AUC). AUC value> 0.6 will be used in the landslide susceptibility zonation next analysis. Validation of landslide susceptibility zonation was carried out using 172 landslide events. The result of the WoE validation shows the AUC success rate of 0.70 and AUC prediction rate of 0.76. The value of AUC shows that the modelling is good and acceptable.

Evaluating methods for debris-flow prediction based on rainfall in an Alpine catchment

Abstract. The prediction of debris flows is relevant because this type of natural hazard can pose a threat to humans and infrastructure. Debris-flow (and landslide) early warning systems often rely on rainfall intensity–duration (ID) thresholds. Multiple competing methods exist for the determination of such ID thresholds but have not been objectively and thoroughly compared at multiple scales, and a validation and uncertainty assessment is often missing in their formulation. As a consequence, updating, interpreting, generalizing and comparing rainfall thresholds is challenging. Using a 17-year record of rainfall and 67 debris flows in a Swiss Alpine catchment (Illgraben), we determined ID thresholds and associated uncertainties as a function of record duration. Furthermore, we compared two methods for rainfall definition based on linear regression and/or true-skill-statistic maximization. The main difference between these approaches and the well-known frequentist method is that non-triggering rainfall events were also considered for obtaining ID-threshold parameters. Depending on the method applied, the ID-threshold parameters and their uncertainties differed significantly. We found that 25 debris flows are sufficient to constrain uncertainties in ID-threshold parameters to ±30 % for our study site. We further demonstrated the change in predictive performance of the two methods if a regional landslide data set with a regional rainfall product was used instead of a local one with local rainfall measurements. Hence, an important finding is that the ideal method for ID-threshold determination depends on the available landslide and rainfall data sets. Furthermore, for the local data set we tested if the ID-threshold performance can be increased by considering other rainfall properties (e.g. antecedent rainfall, maximum intensity) in a multivariate statistical learning algorithm based on decision trees (random forest). The highest predictive power was reached when the peak 30 min rainfall intensity was added to the ID variables, while no improvement was achieved by considering antecedent rainfall for debris-flow predictions in Illgraben. Although the increase in predictive performance with the random forest model over the classical ID threshold was small, such a framework could be valuable for future studies if more predictors are available from measured or modelled data.

Mechanism and Stability Analysis of Deformation Failure of a Slope

Force distribution during progressive slope failure is an important element in slope stability analysis. In this study, five mechanical failure modes are proposed for thrust- and pull-type slopes, respectively, and five field forms of thrust-type slopes are described. The properties of progressive failure are evaluated quantitatively: the failure mode of slope obeys the geo-material rule under the peak stress state, and the instability range is gradually developed. The critical stress state zone is in the process of dynamic change with the development of deformation. It appears that the driving sliding force is greater than the frictional resistance along the sliding surface. When rock or soil stabilizing stresses are at maximum, the vector sum of the driving sliding stress and stabilizing stress is equal to zero at the critical state. The frictional resistance is equal to the driving sliding force in the stable and less-stable regions, and the normal pressure is wherever equal to the counterpressure. Rigid, flexible, and rigid-flexible design theories are proposed for slope control. New terms are defined and used to evaluate the stability. The conventional local and surplus stability factors of slopes and their calculation are explained. The force distribution rule is analyzed during progressive failure, and the conventional stability factor definition is discussed. The geological settings and monitoring data of landslides are used to analyse changes in the critical stress state. An example is given to illustrate the failure process analysis. The results show that progressive failure can be well represented and the safety factor can be well described by the main thrust method (MTM), comprehensive displacement method (CDM), and surplus displacement method (SDM), which can be used to feasibly evaluate slope stability.

Landslide Susceptibility Mapping of Urban Areas: Logistic Regression and Sensitivity Analysis applied to Quito, Ecuador

Although the Andean region is one of the most landslide-susceptible areas in the world, limited attention has been devoted to the topic in this context in terms of research, risk reduction practice, and urban policy. Based on the collection of landslides data of the Andean city of Quito, Ecuador, this article aims to explore the predictive power of a binary logistic regression model (LOGIT) to test secondary data and an official multicriteria evaluation model for landslide susceptibility in this urban area. Cell size resampling scenarios were explored as a parameter, as the inclusion of new “urban” factors. Furthermore, two types of sensitivity analysis (SA), univariate and Monte Carlo methods, were applied to improve the calibration of the LOGIT model. A Kolmogorov–Smirnov (K-S) test was included to measure the classification power of the models. Charts of the three SA methods helped to visualize the sensitivity of factors in the models. The Area Under the Curve (AUC) was a common metric for validation in this research. Among the ten factors included in the model to help explain landslide susceptibility in the context of Quito, results showed that population and street/road density, as novel “urban factors”, have relevant predicting power for landslide susceptibility in urban areas when adopting data standardization based on weights assigned by experts. The LOGIT was validated with an AUC of 0.79. Sensitivity analyses suggested that calibrations of the best-performance reference model would improve its AUC by up to 0.53%. Further experimentation regarding other methods of data pre-processing and a finer level of disaggregation of input data are suggested. In terms of policy design, the LOGIT model coefficient values suggest the need for a deep analysis of the impacts of urban features, such as population, road density, building footprint, and floor area, at a household scale, on the generation of landslide susceptibility in Andean cities such as Quito. This would help improve the zoning for landslide risk reduction, considering the safety, social and economic impacts that this practice may produce.

Large-Scale Landslide Susceptibility Mapping Using an Integrated Machine Learning Model: A Case Study in the Lvliang Mountains of China

Integration of different models may improve the performance of landslide susceptibility assessment, but few studies have tested it. The present study aims at exploring the way to integrating different models and comparing the results among integrated and individual models. Our objective is to answer this question: Will the integrated model have higher accuracy compared with individual model? The Lvliang mountains area, a landslide-prone area in China, was taken as the study area, and ten factors were considered in the influencing factors system. Three basic machine learning models (the back propagation (BP), support vector machine (SVM), and random forest (RF) models) were integrated by an objective function where the weight coefficients among different models were computed by the gray wolf optimization (GWO) algorithm. 80 and 20% of the landslide data were randomly selected as the training and testing samples, respectively, and different landslide susceptibility maps were generated based on the GIS platform. The results illustrated that the accuracy expressed by the area under the receiver operating characteristic curve (AUC) of the BP-SVM-RF integrated model was the highest (0.7898), which was better than that of the BP (0.6929), SVM (0.6582), RF (0.7258), BP-SVM (0.7360), BP-RF (0.7569), and SVM-RF models (0.7298). The experimental results authenticated the effectiveness of the BP-SVM-RF method, which can be a reliable model for the regional landslide susceptibility assessment of the study area. Moreover, the proposed procedure can be a good option to integrate different models to seek an “optimal” result.

Counteracting flawed landslide data in statistically based landslide susceptibility modelling for very large areas: a national-scale assessment for Austria

The reliability of input data to be used within statistically based landslide susceptibility models usually determines the quality of the resulting maps. For very large territories, landslide susceptibility assessments are commonly built upon spatially incomplete and positionally inaccurate landslide information. The unavailability of flawless input data is contrasted by the need to identify landslide-prone terrain at such spatial scales. Instead of simply ignoring errors in the landslide data, we argue that modellers have to explicitly adopt their modelling design to avoid misleading results. This study examined different modelling strategies to reduce undesirable effects of error-prone landslide inventory data, namely systematic spatial incompleteness and positional inaccuracies. For this purpose, the Austrian territory with its abundant but heterogeneous landslide data was selected as a study site. Conventional modelling practices were compared with alternative modelling designs to elucidate whether an active counterbalancing of flawed landslide information can improve the modelling results. In this context, we compared widely applied logistic regression with an approach that allows minimizing the effects of heterogeneously complete landslide information (i.e. mixed-effects logistic regression). The challenge of positionally inaccurate landslide samples was tackled by elaborating and comparing the models for different terrain representations, namely grid cells, and slope units. The results showed that conventional logistic regression tended to reproduce incompleteness inherent in landslide training data in case the underlying model relied on explanatory variables directly related to the data bias. The adoption of a mixed-effects modelling approach appeared to reduce these undesired effects and led to geomorphologically more coherent spatial predictions. As a consequence of their larger spatial extent, the slope unit–based models were able to better cope with positional inaccuracies of the landslide data compared to their grid-based equals. The presented research demonstrates that in the context of very large area susceptibility modelling (i) ignoring flaws in available landslide data can lead to geomorphically incoherent results despite an apparent high statistical performance and that (ii) landslide data imperfections can actively be diminished by adjusting the research design according to the respective input data imperfections.

Rapid Terrain Assessment for Earthquake-Triggered Landslide Susceptibility With High-Resolution DEM and Critical Acceleration

Earthquake ground motion often triggers landslides in mountainous areas. A simple, robust method to quickly evaluate the terrain’s susceptibility of specific locations to earthquake-triggered landslides is important for planning field reconnaissance and rescues after earthquakes. Different approaches have been used to estimate coseismic landslide susceptibility using Newmark’s sliding block model. This model requires an estimate of the landslide depth or thickness, which is a difficult parameter to estimate. We illustrate the use of Newmark sliding block’s critical acceleration for a glaciated valley affected by the 2015 Gorkha earthquake in Nepal. The landslide data came from comparing high-resolution pre- and post-earthquake digital elevation models (DEMs) derived from Spot 6/7 images. The areas where changes were detected provided an inventory of all the landslides triggered by the earthquake. The landslide susceptibility was modeled in a GIS environment using as inputs the pre-earthquake terrain and slope angles, the peak ground acceleration from the 2015 Gorkha earthquake, and a geological map. We exploit the depth information for the landslides (obtained by DEM difference) to apply the critical acceleration model. The spatial distribution of the predicted earthquake-triggered landslides matched the actual landslides when the assumed landslide thickness in the model is close to the median value of the actual landslide thickness (2.6 m in this case). The landslide predictions generated a map of landslide locations close to those observed and demonstrated the applicability of critical acceleration for rapidly creating a map of earthquake-triggered landslides.

Probing the Crisis of Regional Connectivity Instigated by the Natural Disasters, Mizoram, India

Mizoram, in eastern Himalayas of India, is educated but thinly populous young state of area 21081sq km with hilly forest cover of (≈90.68%) The major connectivity is by roads of length 8912km (2019) that connects India’s mainland, Myanmar and Bangladesh and about no rail tracks. The state is housed in various folds and faults over Indo-Burmese tectonic zone (Indo Eurasian plate tectonics), and housed adjacent to 950 Ridges. The booming craggy state is combatting with its road network progress due to frequent low magnitude tremors, landslides, floods, and heavy rainfalls.
 The data of earthquakes, landslides, rainfall, and road progress were collected, and analyzed. The connectivity growth is reported tough and challenging within forests, broken relief, frequent ghats roads, and deep gorges, in topography that hinders the progress in pavement growth. Present study focuses on the ameliorative measures of hill road construction in Mizoram by adhering to the IRC: 52-2019, changing modus operandi of national contract bids, amateurish engineering design, procurement policies, bidder’s expertise and commitment for execution.
 The non-availability menial labours and technocrats have stemmed many legal issues & disputes between the government and the contractors. Faulty planning, engineering design, executant’s approach, procurement policies and the contractor’s skill, contract administration and construction management, road safety challenges, contract documentation, and inability of deployed bidders needs to be improved in Mizoram. To cater the needs of up-gradation/ new constructions, the hurdles warrant cutting edge impulses through EIA, EMP, DPR, amateurish design, and updating procurement policies that should supersede the orthodox practices.

GIS-Based Landslide Susceptibility Mapping using Logistic Regression, Instability Index, and Support Vector Machine: Case Study of the Jingshan River, Taiwan

Many techniques have been developed to produce landslide susceptibility maps. However, mapping the spatial distribution of landslide susceptibility involves critical steps. This study evaluated susceptibility to the occurrence of shallow landslides in upstream areas of the Jingshan River, Taiwan, where heavy rainfall has resulted in increased landslide occurrence and reservoir sedimentation. The landslide susceptibility case study was conducted using logistic regression, an instability index method, and support vector machine (SVM). A selection procedure was first developed to identify the factors influencing landslide occurrence. Historical landslide data were used to assess the corresponding parameters of each model. The receiver operating characteristic curve was then used to evaluate the accuracy of model results. The results indicated that the instability index method underestimated landslide susceptibilities in areas near the river. The instability index method was highly affected by the classifications of model factors. Of the models, the SVR model was the most accurate regarding landslide susceptibility in the study area. Employing this framework, governments may implement suitable, cost-effective improvements to structural measures and land-use planning.

Temporal Variations in Landslide Distributions Following Extreme Events: Implications for Landslide Susceptibility Modeling

AbstractLandslide susceptibility models are fundamental components of landslide risk management strategies. These models typically assume that landslide occurrence is time‐independent, even though processes including earthquake preconditioning and landslide path dependency transiently impact landslide occurrence. Understanding the temporal characteristics of landslide occurrence remains limited by a lack of systematic investigation into how landslide distributions vary through time, and how this impacts landslide susceptibility. Here, we apply Kolmogorov‐Smirnoff and chi‐square statistics to a 30‐yr inventory of monsoon‐triggered landslides from Nepal to systematically quantify how landslide spatial distributions vary through time in “normal” years and years impacted by extreme events. We then develop binary logistic regression (BLR) susceptibility models for 12 yrs in our inventory with >400 landslides and use area under receiver operator curve validation to assess how well these models can hindcast landslide occurrence in other years. Landslide distributions are found to vary through time, particularly in years impacted by storms (1993 and 2002), earthquakes (2015), and floods (2017). Notably, Gorkha earthquake landscape preconditioning shifted 2015 monsoon‐triggered landslides to higher slopes, reliefs, and excess topographies. These variations significantly impact BLR susceptibility modeling, with models trained on extreme years unable to consistently hindcast landslide occurrence in other years. However, developing BLR models using increasingly long historical inventories shows that susceptibility models developed using >6–8 yrs of landslide data provide consistently good hindcasting accuracy. Overall, our results challenge time‐independent assumptions of landslide susceptibility approaches, highlighting the need for time‐dependent modeling techniques or historical inventories for landslide susceptibility modeling.

Analysis of Landslide Susceptibility Using Deep Neural Network

Landslides in Korea are caused by various factors, such as topographic characteristics, geology, and climate change, and they cause significant damage to property and human life. It is necessary to analyze landslide susceptibility to identify the location of landslide occurrence precisely and respond to the risk of landslides. In this study, the probability of landslide occurrence was calculated through a landslide sensitivity analysis using a deep neural network based on eight conditioning factors and 26 landslide data. In addition, verification was performed using the ROC method. The landslide susceptibility obtained using a deep neural network showed a success rate of 70% and a prediction rate of 81.7%, indicating that the prediction rate was 11.7% higher than the success rate. In addition, a landslide susceptibility map for estimating the probability of landslide occurrence was plotted using the geometric spacing method. The chi-square test results indicated that the landslide susceptibility map obtained in this study was statistically significant. The location of landslides can be identified more accurately using the proposed method.

National-scale data-driven rainfall induced landslide susceptibility mapping for China by accounting for incomplete landslide data

China is one of the countries where landslides caused the most fatalities in the last decades. The threat that landslide disasters pose to people might even be greater in the future, due to climate change and the increasing urbanization of mountainous areas. A reliable national-scale rainfall induced landslide susceptibility model is therefore of great relevance in order to identify regions more and less prone to landsliding as well as to develop suitable risk mitigating strategies. However, relying on imperfect landslide data is inevitable when modelling landslide susceptibility for such a large research area. The purpose of this study is to investigate the influence of incomplete landslide data on national scale statistical landslide susceptibility modeling for China. In this context, it is aimed to explore the benefit of mixed effects modelling to counterbalance associated bias propagations. Six influencing factors including lithology, slope, soil moisture index, mean annual precipitation, land use and geological environment regions were selected based on an initial exploratory data analysis. Three sets of influencing variables were designed to represent different solutions to deal with spatially incomplete landslide information: Set 1 (disregards the presence of incomplete landslide information), Set 2 (excludes factors related to the incompleteness of landslide data), Set 3 (accounts for factors related to the incompleteness via random effects). The variable sets were then introduced in a generalized additive model (GAM: Set 1 and Set 2) and a generalized additive mixed effect model (GAMM: Set 3) to establish three national-scale statistical landslide susceptibility models: models 1, 2 and 3. The models were evaluated using the area under the receiver operating characteristics curve (AUROC) given by spatially explicit and non-spatial cross-validation. The spatial prediction pattern produced by the models were also investigated. The results show that the landslide inventory incompleteness had a substantial impact on the outcomes of the statistical landslide susceptibility models. The cross-validation results provided evidence that the three established models performed well to predict model-independent landslide information with median AUROCs ranging from 0.8 to 0.9. However, although Model 1 reached the highest AUROCs within non-spatial cross-validation (median of 0.9), it was not associated with the most plausible representation of landslide susceptibility. The Model 1 modelling results were inconsistent with geomorphological process knowledge and reflected a large extent the underlying data bias. The Model 2 susceptibility maps provided a less biased picture of landslide susceptibility. However, a lower predicted likelihood of landslide occurrence still existed in areas known to be underrepresented in terms of landslide data (e.g., the Kuenlun Mountains in the northern Tibetan Plateau). The non-linear mixed-effects model (Model 3) reduced the impact of these biases best by introducing bias-describing variables as random effects. Among the three models, Model 3 was selected as the best national-scale susceptibility model for China as it produced the most plausible portray of rainfall induced landslide susceptibility and the highest spatially explicit predictive performance (median AUROC of spatial cross validation 0.84) compared to the other two models (median AUROCs of 0.81 and 0.79, respectively). We conclude that ignoring landslide inventory-based incompleteness can entail misleading modelling results and that the application of non-linear mixed-effect models can reduce the propagation of such biases into the final results for very large areas.

Comparing methods for determining landslide early warning thresholds: potential use of non-triggering rainfall for locations with scarce landslide data availability

Rainfall intensity-duration landslide-triggering thresholds have become widespread for the development of landslide early warning systems. Thresholds can be in principle determined using rainfall event datasets of three types: (a) rainfall events associated with landslides (triggering rainfall) only, (b) rainfall events not associated with landslides (non-triggering rainfall) only, (c) both triggering and non-triggering rainfall. In this paper, through Monte Carlo simulation, we compare these three possible approaches based on the following statistical properties: robustness, sampling variation, and performance. It is found that methods based only on triggering rainfall can be the worst with respect to those three investigated properties. Methods based on both triggering and non-triggering rainfall perform the best, as they could be built to provide the best trade-off between correct and wrong predictions; they are also robust, but still require a quite large sample to sufficiently limit the sampling variation of the threshold parameters. On the other side, methods based on non-triggering rainfall only, which are mostly overlooked in the literature, imply good robustness and low sampling variation, and performances that can often be acceptable and better than thresholds derived from only triggering events. To use solely triggering rainfall—which is the most common practice in the literature—yields to thresholds with the worse statistical properties, except when there is a clear separation between triggering and non-triggering events. Based on these results, it can be stated that methods based only on non-triggering rainfall deserve wider attention. Methods for threshold identification based on only non-triggering rainfall may have the practical advantage that can be in principle used where limited information on landslide occurrence is available (newly instrumented areas). The fact that relatively large samples (about 200 landslides events) are needed for a sufficiently precise estimation of threshold parameters when using triggering rainfall suggests that threshold determination in future applications may start from identifying thresholds from non-triggering events only, and then move to methods considering also the triggering events as landslide information starts to become more available.

Landslides susceptibility mapping based on geospatial data and geomorphic attributes (a case study: Pacet, Mojokerto, East Java)

The study of landslides for long time is studied by geomorphological approach. In the study of geomorphology, each and every spatial segment of the earth surface possesses some physiographic aspects. Its analysis enables us to predict an interrelationship between physical and cultural phenomena as a whole. Pacet is one of the most susceptible areas due to landslides in Mojokerto Regency, East Java. Pacet is located at mountainous area which the slope stability depends upon physical and chemical properties of the soil. Such condition is represented on the geomorphic properties, such as slope angle, slope aspect, profile curvature, TWI, TPI, SPI and lithological composition. To obtain a proper landslides susceptibility mapping, some data were derived from geospatial data such as slope angle, slope aspect, profile curvature, TWI, TPI and SPI. DEM extraction by GIS platform were used to obtain the data. Soil samples were collected from different landform unit. The spatial distribution of landslides data was processed using GIS Software. The result shows that landslides were influenced by slope, aspect, profile curvature, TWI, TPI, SPI and lithological composition in study area.

GIS-based comparative study of Bayes network, Hoeffding tree and logistic model tree for landslide susceptibility modeling

Landslides, one of the most common hazards around the world, have brought about severe damage to life and property of human. To prevent and mitigate landslides, various models have been introduced to assess landslide susceptibility. In this paper, Hoeffding Tree (HT), a prevailing data stream mining algorithm, was employed to predict landslide susceptibility in Muchuan County, China for the first time. Meanwhile, Logistic Model Tree (LMT) and Bayes Network (BN) were applied to produce landslide susceptibility maps for comparison. The model performances were evaluated by Receiver Operating Characteristic (ROC) curves and areas under the curves (AUC). To obtain landslide inventory map, 279 landslides data was collected, and training and validation datasets were randomly divided with a proportion of 70% to 30%. Furthermore, twelve conditioning factors (altitude, slope angle, profile curvature, plan curvature, slope aspect, distance to roads, distance to rivers, TWI, NDVI, soil, land use and lithology) were selected to construct landslide susceptibility models. Moreover, correlations between conditioning factors and landslides were analyzed using Frequency Ratio (FR). The results showed landslides are prone to occur in areas where human activities concentrate, and all three models exhibited satisfying performances. Concretely, for training dataset, LMT model showed the highest AUC (0.854), followed by HT (0.726) and BN (0.709). However, for validation dataset, LMT and BN models generated similar AUC values (0.761 and 0.764 respectively), and the highest AUC value belonged to HT (0.802). The distributions of landslide susceptibility zones revealed that the interior of county town is mainly seated in low and very low susceptibility zones, whereas regions close to the border suffer high and very high landslide risk. The results acquired in this paper are significant to landslide prevention and urban planning in Muchuan, China. Additionally, this study proved that HT model is a promising classifier for landslide susceptibility modeling.

Derivation of earthquake-induced landslide distribution using aerial photogrammetry: the January 24, 2020, Elazig (Turkey) earthquake

On January 24, 2020, an earthquake with the Mw of 6.8 occurred on the East Anatolian Fault Zone in Elazig Province, Turkey, and triggered many landslides. Even though Turkey is in a high seismic zone and has highly susceptible areas to landslides, event-based inventories of landslides triggered by earthquakes have not yet been published in the international literature. The purpose of the present study is to compile the landslide inventory triggered by the Elazig earthquake. A novel approach involved change detection analyses was applied by using high-resolution 3D digital surface models (DSMs) obtained from pre- and post-earthquake aerial photos. In the study, a total of 328 landslides sized between 133 m2 and 3 × 106 m2 were mapped by visual interpretations. The rollover effect and fractal dimension of the failures were obtained as 3600 m2 and –1.48, respectively. The total landslide area including new active zones developed after 2018 within the existing mass was calculated as approximately 8 × 106 m2. According to the volumetric change analyses, the displacements were between –17.66 m and 14.11 m. It is evident that using photogrammetric methods to produce very high-resolution DSMs enables precise determination of landslide activity after an earthquake and increases the quality of the inventory. Consequently, application of the surface comparison methods immediately after large earthquakes provides great benefits in obtaining new landslide data open up important possibilities for a better understanding of the mechanism of landslides triggered by earthquakes, thus minimizing losses sourced from them.

Spatio-temporal variability of monsoon precipitation and their effect on precipitation triggered landslides in relation to relief in Himalayas

Orography of Himalaya governs the regional weather system and monsoon of Indian sub-continent. The intense persistent precipitation in the form of rainfall during monsoon season causes landslides which are the most frequent naturally occurring hazards in the Himalaya. This study attempts to investigate the spatio-temporal variability of precipitation and their effect on precipitation triggered landslides during monsoon season (June–September) in relation to relief in Himalayan region by utilizing satellite-derived precipitation products (GPM) over the span of 2000–2018 in addition to NASA Landslide Viewer, Global Landslide Catalog and Global Risk Data Platform data for landslides and ASTER DEM for elevation. The results exhibit that the Eastern Himalayas received the highest amount of precipitation of 2385 mm with intensity of 19.5 mm/day, followed by the Central Himalayas with 1860 mm and intensity of 17.5 mm/day and the least in the Western Himalayas with 1400 mm and intensity of 15 mm/day, respectively. The monsoon precipitation in the lower elevations (below 2600 m) are mostly responsible for causing a vast majority of the precipitation induced landslide events with a maximum of 68.66% in the Central Himalayas followed by the Western Himalayas at 62.23% and the least in the Eastern Himalayas at 41.16%. The overall accumulated precipitation and intensity during monsoon season and landslide distribution were strongly correlated with relief pattern. This comprehensive study signifies how relief regulated the occurrences of landslides in monsoon season and recommends vegetation cover and least interference of human-induced land use to alleviate the landslides.

Model performance analysis for landslide susceptibility in cold regions using accuracy rate and fluctuation characteristics

Considering the increasing number of landslides due to permafrost degradation, this paper reports a performance evaluation of three classical landslide susceptibility models applied to cold regions. A landslide inventory was first constructed through a historical survey and image interpretation. Ten causative factors of landslides were then chosen based on the available data and the local environment. Multicollinearity diagnosis and factor effectiveness test were employed to perform a factor analysis. Subsequently, three evaluation models based on the frequency ratio (FR), logistic regression (LR), and artificial neural network (ANN) were established. These models were applied to obtain landslide susceptibility maps, which were then analyzed and compared. The model performance was evaluated in terms of the accuracy rate and fluctuation characteristics. The results showed no multicollinearity issue between the factors employed. The annual temperature difference and frozen depth are two indispensable factors when assessing landslide susceptibility in cold regions. A comparison between the susceptibility maps generated using the three models showed that the FR model-generated map is most in line with the principle of disaster zoning and has the highest degree of conformity with actual landslide points, followed by the maps generated using the LR and ANN models. An accuracy analysis showed that the ANN model yields the highest AUC value in the training and test states, 0.957 and 0.863, respectively; however, these values were not optimal given the fluctuation. Moreover, the fluctuation in the non-landslide data was greater than that in the landslide data. The fluctuation results revealed the drawback of the AUC value in the analysis of the model performance. In other words, the non-landslide error often covers up the landslide error. This study provides a scientific guidance for evaluating the model performance and for assessing landslide disasters in cold regions.

Assessment of Submarine Landslide Susceptibility in the Sea Area of Zhoushan

The research on the submarine landslide susceptibility has important guiding significance for the site selection and construction of marine engineering. The Zhoushan Archipelago New Area is dominated by the marine industry, and the marine engineering under its jurisdiction is densely covered. This research selected the surrounding sea area of Zhoushan island as the study area to evaluate submarine landslide susceptibility. The factors such as seismic peak acceleration, proximity to fault zone, average wave height, slope and sediment types were selected to construct the evaluation index system, all of which were considered to be influential in the occurrence of landslides. Meanwhile, the AHP method was used to calculate the weight of each index factor. In combination with geographical information system, the study area has been classified into five classes of relative landslide susceptibility, namely, extremely low, low, medium, medium-high and high. The research result showed that: (1) The high susceptibility zone in the evaluation result was consistent with the position of historical landslide. (2) The five classes in the result of submarine landslide susceptibility assessment for study area accounted for 1.17%, 1.97%, 9.08%, 56.05% and 31.74% of the total area from high to low, and the area with susceptibility above middle grade mainly distributed in the underwater bank slope around the islands or the slopes position of the tidal current scours. (3) This research was a preliminary study on the susceptibility assessment of regional submarine landslides. It was limited by the availability of evaluation index data and submarine landslide data accumulation in the region. Besides that, the evaluation model was influenced by experience greatly, so the evaluation result had certain subjectivity.