Search

Reservoir impoundment is the main factor inducing reservoir landslides, it is essential to obtain the dynamic landslide susceptibility by considering hydraulic factors. In this paper, we proposed the use of the original and revised logistic regression models (machine learning methods) to analyze the landslide susceptibility after the second stage of reservoir impoundment. Data of active unstable slopes before the first stage of reservoir impoundment were used to train the models. The results indicated that the two models both performed well with high accuracy. The revised logistic regression model performed better than the original logistic regression model in two aspects. Firstly, TPR and AUC of the revised model is 0.771 and 0.906 which is obviously higher than the original model of 0.743 and 0.896. Secondly, the revised model can predict the unstable slopes after reservoir impoundment which the original model can’t predict. The former could be used to obtain the dynamic landslide susceptibility in reservoir areas after different stages of reservoir impoundment and predict the potential reactivated unstable slopes more precisely due to hydraulic factors.

The Karakoram Highway (KKH) is a major road traffic route connecting China to South Asia and the Middle East via Pakistan. Geological hazards occur often in the study area due to the influence of complex geological features, active glaciers, and a changing climate, causing major disruptions to traffic operations. As a result, a geological hazard susceptibility map was created in this study to aid in the prevention and control of highway slope disasters. Through field investigation and map interpretation, 102 geohazards, including landslides, debris flows, and collapses, were observed along the route. The logistic regression method was utilized to generate a predictive model for the susceptibility map considering topography, faults, peak ground acceleration, geology, land cover, rainfall, and drainages. The results imply that the model performs well in terms of prediction. The AUC (area under the receiver operating characteristics (ROC) curve) value is 90.6%. The risk associated with the highway was assessed using the susceptibility map. Some sections of the highway are in high-risk zones, where disasters are more perilous and endanger people and automobiles. This effort will support geohazard protection, management, and safe operation of road traffic.

The improvement of landslide susceptibility assessment is a long-standing problem in hazard mitigation work, wherein previous studies have proposed various training models. However, the ratio of positive to negative samples and the selection of non-landslide samples have been shown to significantly influence results. These research directions have traditionally been focal points, while datasets are often overlooked, serving merely as auxiliary tools to support the validation process. Hence, this study proposes an approach to enhance datasets through the introduction of the side-sampling method. This technique focuses on individual research cells, conducting feature sampling training on fixed regions of length M, thereby enabling more precise identification of geographical clustering characteristics. Using evaluation metrics such as accuracy, precision, recall, F1 score, and ROC curve, this study conducts a comparative analysis between the side-sampling method and traditional sampling methods, using three distinct railway lines in China as the study areas. Results show substantial improvements beyond several exceptions: accuracy (+7.68%), precision (+7.19%), recall (+13.48%), F1 score (+9.92%), and ROC (+6.22%). The results demonstrate a significant overall improvement in the performance of the trained models based on the side-sampling method, providing a positive insight into mitigating landslide hazards along railways from the dataset perspective.

Strong earthquakes can frequently trigger a large number of co-seismic landslides. Obtaining an accurate susceptibility map of co-seismic landslides is crucial for post-disaster rescue and reconstruction efforts. In this study, the pre-seismic average annual surface deformation rate of the study area was obtained using Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) technology. Furthermore, the landslide hazards prior to the earthquake in multiple locations within the study area were analyzed. Subsequently, the deformation data was combined with 11 evaluation factors, including the distance to the fault and the Peak Ground Acceleration (PGA), to model the susceptibility of landslides. We constructed four models to evaluate the susceptibility of landslides in the study area: the Bayesian optimization random forest (BO-RF) model, the random forest model (RF), the logistic regression model (LR), and the support vector machine model (SVM). The BO-RF model outperformed the other models, achieving an AUC value of 0.984, an accuracy of 0.952, a precision of 0.953, a recall of 0.952, and an F1 score of 0.953. Furthermore, incorporating pre-seismic deformation features in the evaluation of co-seismic landslide susceptibility can effectively improve the reliability of model predictions, as compared to the evaluation model results without incorporating deformation factors. The obtained research results provide valuable data support for the rescue and management of disaster-stricken areas.

A landslide susceptibility map, which describes the quantitative relationship between known landslides and control factors, is essential to link the theoretical prediction with practical disaster reduction measures. In this work, the artificial neural network (ANN) model, a promising tool for mapping landslide susceptibility, was adopted to evaluate the coseismic landslide susceptibility affected by the 2013 Minxian, Gansu, China, Mw5.9 earthquake. The evaluation was based on the landslide inventory of this event containing 6479 landslides, and the terrain, geological and seismic factors from database available. During the analyses, two ANN models were applied: considering the entire factors aforementioned (CS model) and excluding seismic factors above (ES model). The success and predictive rates of ANN models and the cumulative percentage curves of susceptibility maps obtained from the models all indicate that the CS model has a relatively better performance than the ES model. However, the comparison of overlapping susceptibility areas suggests that 52.8% of the very high susceptibility areas derived from the CS model coincide with the ES model; and for the very low susceptibility areas, this proportion is 73.55%. Thus, it can be concluded that the assessment based on existing earthquake-induced landslides and the ES model could provide better background information for seismic landslide susceptibility mapping and disaster prevention.

A landslide susceptibility map, which describes the quantitative relationship between known landslides and control factors, is essential to link the theoretical prediction with practical disaster reduction measures. In this work, the artificial neural network (ANN) model, a promising tool for mapping landslide susceptibility, was adopted to evaluate the coseismic landslide susceptibility affected by the 2013 Minxian, Gansu, China, Mw5.9 earthquake. The evaluation was based on the landslide inventory of this event containing 6479 landslides, and the terrain, geological and seismic factors from database available. During the analyses, two ANN models were applied: considering the entire factors aforementioned (CS model) and excluding seismic factors above (ES model). The success and predictive rates of ANN models and the cumulative percentage curves of susceptibility maps obtained from the models all indicate that the CS model has a relatively better performance than the ES model. However, the comparison of overlapping susceptibility areas suggests that 52.8% of the very high susceptibility areas derived from the CS model coincide with the ES model; and for the very low susceptibility areas, this proportion is 73.55%. Thus, it can be concluded that the assessment based on existing earthquake-induced landslides and the ES model could provide better background information for seismic landslide susceptibility mapping and disaster prevention. [ABSTRACT FROM AUTHOR]