Among the most frequent and dangerous natural hazards, landslides often result in huge casualties and economic losses. Landslide susceptibility mapping (LSM) is an excellent approach for protecting and reducing the risks by landslides. This study aims to explore the performance of Bayesian optimization (BO) in the random forest (RF) and gradient boosting decision tree (GBDT) model for LSM and applied in Shuicheng County, China. Multiple data sources are used to obtain 17 conditioning factors of landslides, Borderline-SMOTE and Randomundersample methods are combined to solve the imbalanced sample problem. RF and GBDT models before and after BO are adopted to calculate the susceptibility value of landslides and produce LSMs and these models were compared and evaluated using multiple validation approach. The results demonstrated that the models we proposed all have high enough model accuracy to be applied to produce LSM, the performance of the RF is better than the GBDT model without BO, while after adopting the Bayesian optimized hyperparameters, the prediction accuracy of the RF and GBDT models is improved by 1% and 7%, respectively and the Bayesian optimized GBDT model is the best for LSM in this four models. In summary, the Bayesian optimized RF and GBDT models, especially the GBDT model we proposed for landslide susceptibility assessment and LSM construction has a very good application performance and development prospects.

Landslides pose a constant threat to the lives and property of mountain people and may also cause geomorphological destruction such as soil and water loss, vegetation destruction, and land cover change. Landslide susceptibility assessment (LSA) is a key component of landslide risk evaluation. There are many related studies, but few analyses and comparisons of models for optimization. This paper aims to introduce the Tree-structured Parzen Estimator (TPE) algorithm for hyperparameter optimization of three typical neural network models for LSA in Shuicheng County, China, as an example, and to compare the differences of predictive ability among the models in order to achieve higher application performance. First, 17 influencing factors of landslide multiple data sources were selected for spatial prediction, hybrid ensemble oversampling and undersampling techniques were used to address the imbalanced sample and small sample size problem, and the samples were randomly divided into a training set and validation set. Second, deep neural network (DNN), recurrent neural network (RNN), and convolutional neural network (CNN) models were adopted to predict the regional landslides susceptibility, and the TPE algorithm was used to optimize the hyperparameters respectively to improve the assessment capacity. Finally, to compare the differences and optimization effects of these models, several objective measures were applied for validation. The results show that the high-susceptibility regions mostly distributed in bands along fault zones, where the lithology is mostly claystone, sandstone, and basalt. The DNN, RNN, and CNN models all perform well in LSA, especially the RNN model. The TPE optimization significantly improves the accuracy of the DNN and CNN (3.92% and 1.52%, respectively), but does not improve the performance of the RNN. In summary, our proposed RNN model and TPE-optimized DNN and CNN model have robust predictive capability for landslide susceptibility in the study area and can also be applied to other areas containing similar geological conditions.

Landslides are among the most frequent natural hazards in the world. Rainfall is an important triggering factor for landslides and is responsible for topples, slides, and debris flows—three of the most important types of landslides. However, several previous relevant research studies covered general landslides and neglected the rainfall–topples–slides–debris flows disaster chain. Since landslide hazard mapping (LHM) is a critical tool for disaster prevention and mitigation, this study aimed to build a GeoDetector and Bayesian network (BN) model framework for LHM in Shuicheng County, China, to address these geohazards. The GeoDetector model will be used to screen factors, eliminate redundant information, and discuss the interaction between elements, while the BN model will be used for constructing a causality disaster chain network to determine the probability and risk level of the three types of landslides. The practicability of the BN model was confirmed by error rate and scoring rules validation. The prediction accuracy results were tested using overall accuracy, Matthews correlation coefficient, relative operating characteristics curve, and seed cell area index. The proposed framework is demonstrated to be sufficiently accurate to construct the complex LHM. In summary, the combination of the GeoDetector and BN model is very promising for spatial prediction of landslides.

Tianchi volcano is a dormant active volcano with a risk of re-eruption. Volcanic soil and volcanic ash samples were collected around the volcano and the concentrations of 21 metals (major and trace elements) were determined. The spatial distribution of the metals was obtained by inverse distance weight (IDW) interpolation. The metals’ sources were identified and their pollution levels were assessed to determine their potential ecological and human health risks. The metal concentrations were higher around Tianchi and at the north to the west of the study area. According to the geo-accumulation index (I_geo), enrichment factor (EF) and contamination factor (CF) calculations, Zn pollution was high in the study area. Pearson’s correlation analysis and principal component analysis showed that with the exception of Fe, Mn and As, the metals that were investigated (Al, K, Ca, Na, Mg, Ti, Cu, Pb, Zn, Cr, Ni, Ba, Ga, Li, Co, Cd, Sn, Sr) were mostly naturally derived. A small proportion of Li, Pb and Zn may have come from vehicle traffic. There is no potential ecological risk and non-carcinogenic risk because of the low concentrations of the metals; however, it is necessary to pay attention to the carcinogenic risk of Cr and As in children.

The impacts of earthquakes and secondary disasters on ecosystems and the environment are attracting increasing global attention. Development of uncertainty reasoning models offers a chance to research these complex correlations. The primary aim of this research was to construct a disaster chain hazard assessment model that combines a Bayesian Network model and the ArcGIS program software for Changbai Mountain, China, an active volcano with a spate of reported earthquakes, collapses, and landslide events. Furthermore, the probability obtained by the Bayesian Networks was used to determine the disaster chain probability and hazard intensity of the earthquake events, while ArcGIS was used to produce the disaster chain hazard map. The performance of the Bayesian Network model was measured by error rate and scoring rules. The confirmation of the outcomes of the disaster chain hazard assessment model shows that the model demonstrated good predictive performance on the basis of the area under the curve, which was 0.7929. From visual inspection of the produced earthquake disaster chain hazard map, highly hazardous zones are located within a 15 km radius from the Tianchi center, while the northern and the western parts of the studied area are characterized mainly by “very low” to “low” hazard values.

Severe natural disasters and related secondary disasters are a huge menace to society. Currently, it is difficult to identify risk formation mechanisms and quantitatively evaluate the risks associated with disaster chains; thus, there is a need to further develop relevant risk assessment methods. In this research, we propose an earthquake disaster chain risk evaluation method that couples Bayesian network and Newmark models that are based on natural hazard risk formation theory with the aim of identifying the influence of earthquake disaster chains. This new method effectively considers two risk elements: hazard and vulnerability, and hazard analysis, which includes chain probability analysis and hazard intensity analysis. The chain probability of adjacent disasters was obtained from the Bayesian network model, and the permanent displacement that was applied to represent the potential hazard intensity was calculated by the Newmark model. To validate the method, the Changbai Mountain volcano earthquake–collapse–landslide disaster chain was selected as a case study. The risk assessment results showed that the high-and medium-risk zones were predominantly located within a 10 km radius of Tianchi, and that other regions within the study area were mainly associated with very low-to low-risk values. The verified results of the reported method showed that the area of the receiver operating characteristic (ROC) curve was 0.817, which indicates that the method is very effective for earthquake disaster chain risk recognition and assessment.