Article
Geosciences, Multidisciplinary
Sixiang Ling, Siyuan Zhao, Junpeng Huang, Xuantu Zhang
Summary: This study compared the prediction ability of statistical and machine learning models in landslide susceptibility assessment. The results showed that the machine learning models are superior to statistical models in generating adequate landslide susceptibility maps, with the LMT model being the most efficient for landslide prediction in the study region.
FRONTIERS IN EARTH SCIENCE
(2022)
Article
Engineering, Environmental
Yifan Sheng, Yuanyao Li, Guangli Xu, Zhigang Li
Summary: This paper proposes a multidimensional landslide warning method based on statistical and physical models. The method establishes threshold models for cumulative rainfall-duration-mean intensity based on landslide events in a specific region. The thresholds are compared using different evaluation indicators, and the most suitable threshold equation is determined. The method is further applied to analyze slope stability under different rainfall return periods, and the rainfall threshold inducing landslides is determined. The appropriateness of the threshold equation is verified through the analysis of landslide events, and the results are compared with similar working areas globally. This method provides a theoretical basis for preventing and managing landslide disasters and has significant academic value and practical implications.
BULLETIN OF ENGINEERING GEOLOGY AND THE ENVIRONMENT
(2022)
Article
Environmental Sciences
Nafiseh Rezapour Andabili, Mahsa Safaripour
Summary: This study modeled landslide susceptibility in the Miandoab Country of northwestern Iran using the random forest algorithm. The results showed high accuracy of the model and identified digital height layer, geology, and distance from fault as the main factors affecting earthquake potential. The study also classified the study area into low, moderate, and high risk zones and evaluated precipitation trends in the region.
ENVIRONMENTAL MONITORING AND ASSESSMENT
(2022)
Article
Environmental Sciences
Chuanming Ma, Zhiwei Yan, Peng Huang, Lin Gao
Summary: This paper explores the susceptibility characteristics of landslides in Yuan'an County using the AHP-CI model, analyzing both inherent and inducing factors affecting landslide occurrence. By evaluating geological conditions and utilizing GIS, different prevention measures for varying levels of susceptibility are proposed, providing valuable insights for landslide evaluation and prevention efforts in other regions.
ENVIRONMENTAL EARTH SCIENCES
(2021)
Article
Engineering, Geological
Fei Ye, Wen-Xi Fu, Hong-Fu Zhou, Yue Liu, Ren-Ji Ba, Shuang Zheng
Summary: This study investigated a rainfall-induced catastrophic landslide that occurred in Zhonghaicun village, Hanyuan County, Sichuan Province, China on August 21, 2020. The findings revealed the significant impact of long-term intensive rainfall and previous earthquakes on the evolution of the landslide, as well as the presence of multiple landslides and a debris flow in the study area.
Article
Engineering, Multidisciplinary
Zhou Ke, Wang Hong, Jianxing Liao, Yuguang Zhang, Fangping Chen, Zhengjun Yang
Summary: The failure mechanism of Xinzhan landslide was systematically investigated using multiple-integrated geotechniques, including field investigations, monitoring, and numerical simulations. The occurrence of Xinzhan landslide is attributed to its unique geological environment and sustained rainfall. Under sustained rainfall, the infiltration of rainwater into the gravelly soil weakens the strength parameter near the soil-rock interface, leading to intensified plastic deformation. The study also establishes a critical rainfall early warning criterion for Xinzhan landslide based on the rainfall intensity-duration curve.
ALEXANDRIA ENGINEERING JOURNAL
(2023)
Article
Engineering, Geological
Qin Panpan, Huang Bolin, Li Bin, Chen Xiaoting, Jiang Xiannian
Summary: This study successfully reproduced and predicted a landslide event in Guang'an Village, Chongqing, China using a granular flow model and an elasto-visco-plasticity model. The results showed that the landslide gradually moved along the sliding surface, pushing loose deposits and blocking the river. The study also identified multiple deformation areas, with one area of particular concern that may cause the river to be blocked again.
Article
Multidisciplinary Sciences
Maliyamuguli Abulimiti, Zibibula Simayi, Shengtian Yang, Ziyuan Chai, Yibo Yan
Summary: This paper examines the current situation and coordination of urbanization development in Xinjiang using statistical data and various models. The results show that Xinjiang's urbanization level exhibits a center-periphery development pattern and most counties are in a serious state of imbalance.
Article
Green & Sustainable Science & Technology
Minghong Li, Yuanxiang Guo, Danyuan Luo, Chuanming Ma
Summary: In this study, variable weight theory (VW) is introduced to improve the MLP and AHP methods, and two hybrid models, AHP-VW and MLP-VW, are proposed. By redistributing the weights of factors using VW theory, the subjectivity and randomness problems are eliminated. The landslide susceptibility maps of the four models are validated using ROC curve, and the results show that the MLP model hybrid VW performs the best. The maps can be used for regional land use planning and landslide hazard mitigation purposes.
Article
Geosciences, Multidisciplinary
Qigen Lin, Pedro Lima, Stefan Steger, Thomas Glade, Tong Jiang, Jiahui Zhang, Tianxue Liu, Ying Wang
Summary: China is one of the countries with high landslide fatality rates. The reliability of national-scale landslide susceptibility models is crucial for identifying high-risk areas and developing risk mitigation strategies. Incomplete landslide data can impact the accuracy of these models, and non-linear mixed-effect models can help reduce biases in large-scale modeling.
GEOSCIENCE FRONTIERS
(2021)
Article
Engineering, Multidisciplinary
Jianhui Dong, Mao Qiu, Jianjun Zhao, Haijun Li, Qihong Wu
Summary: This study mainly explores the deformation instability mechanism and influencing factors of landslide in Fa'er town, Guizhou Province, southwestern China. The results indicate that the mining of underground coal seams resulted in multiple tensile cracks on the trailing edge. The fragmented rock and soil mass, as well as the free face on the front edge, created favorable landforms for initiating the landslide. The landslide was mainly induced by heavy rainfall for several consecutive days. Under continuous heavy rainfall, the creep shear zone was penetrated by cracks, causing the landslide to slide due to gravity, and the deformation mode shifted from creeping-tensile cracking to plastic flow-tensile cracking.
ALEXANDRIA ENGINEERING JOURNAL
(2022)
Article
Geosciences, Multidisciplinary
Jiang Xiong, Chuan Tang, Ming Chen, Lingfeng Gong, Ning Li, Xianzheng Zhang, Qingyun Shi
Summary: The study showed that the capacity of the landslide sediment supply for channels in debris flow basins generally decreased from 2008 to 2018, but may gradually diminish in an unstable way due to heavy rainfall. The long-term evolution of landslides plays a key role in influencing the sediment supply capacity of landslides in the study area. Higher debris flow activity may last longer as there is still sufficient material accumulated in the channels.
Article
Environmental Sciences
Dongfen Li, Xiaochuan Tang, Zihan Tu, Chengyong Fang, Yuanzhen Ju
Summary: Landslide detection and distribution mapping are crucial for geohazard prevention. In this study, LiDAR data and optical images were used together to develop an automatic detection model called DemDet, which achieved high accuracy, mIoU, and F1 values.
Article
Multidisciplinary Sciences
Xiangpeng Wang, Kunpeng Wang, Fanqiang Lin, Kai Guo
Summary: Based on the analysis of the formation mechanism, stability state, and development trend, this paper focuses on monitoring key indicators such as surface deformation and rainfall to provide early warning and guidance for landslide disaster prevention in the alpine mountainous area of Nangqian, Qinghai Province.
SCIENTIFIC REPORTS
(2022)
Article
Geosciences, Multidisciplinary
Jingjing Jing, Zhijian Wu, Chengxin Chu, Wanpeng Ding, Wei Ma
Summary: The study assessed the hazards of earthquake-induced landslides in Litang County, Sichuan Province, China, considering the amplification effect of site and topography on ground motion parameters and making PGA correction. The results showed that the hazards were higher in high slopes of loose rock and steep rock slopes with large topographic relief. Site and topographic conditions significantly affected the nonlinear amplification of PGA, with corrected PGA being magnified in steep mountains. Rare earthquakes had a greater influence on landslide initiation and movement distance compared to occasional earthquakes. This study provides valuable references for landslide hazard assessment and emergency response in Litang County.
Article
Environmental Sciences
Yunzhi Chen, Wei Chen, Saeid Janizadeh, Gouri Sankar Bhunia, Amit Bera, Quoc Bao Pham, Nguyen Thi Thuy Linh, Abdul-Lateef Balogun, Xiaojing Wang
Summary: This study compared multiple algorithms and found that the Deep Boosting model performed best in piping erosion susceptibility mapping. The results showed that 41% of agricultural lands are very sensitive to piping erosion, aiding resource managers and planners in making effective decisions.
GEOCARTO INTERNATIONAL
(2022)
Article
Environmental Sciences
Yang Li, Wei Chen, Fatemeh Rezaie, Omid Rahmati, Davoud Davoudi Moghaddam, John Tiefenbacher, Mahdi Panahi, Moung-Jin Lee, Dominik Kulakowski, Dieu Tien Bui, Saro Lee
Summary: By combining convolutional neural network with two evolutionary optimization algorithms, debris flow susceptibility maps were successfully generated, pinpointing the factors that most affect debris flow likelihood. The study demonstrates that the models hybridized with optimization algorithms show higher goodness-of-fit and predictive power.
GEOCARTO INTERNATIONAL
(2022)
Article
Environmental Sciences
Yunzhi Chen, Wei Chen, Subodh Chandra Pal, Asish Saha, Indrajit Chowdhuri, Behzad Adeli, Saeid Janizadeh, Adrienn A. Dineva, Xiaojing Wang, Amirhosein Mosavi
Summary: This research focused on delineating potential groundwater zones in Saveh City, Iran, using a hybrid deep learning and machine learning algorithm. The models, including boosted tree (BT), artificial neural network (ANN), deep learning neural network (DLNN), deep learning tree (DLT), and deep boosting (DB), showed good performance in groundwater potential mapping. Factors such as altitude, rainfall, distance to fault, and soil types were identified as important for groundwater potential modeling.
GEOCARTO INTERNATIONAL
(2022)
Article
Environmental Sciences
Xiaojing Wang, Faming Huang, Xuanmei Fan, Himan Shahabi, Ataollah Shirzadi, Huiyuan Bian, Xiongde Ma, Xinxiang Lei, Wei Chen
Summary: This study applies three advanced landslide susceptibility models to evaluate landslide susceptibility in Muchuan County, China. By analyzing the local geo-environmental characteristics and a landslide inventory map, the best model is determined and valuable information for slope stability is provided for local governments and organizations.
ENVIRONMENTAL EARTH SCIENCES
(2022)
Article
Geochemistry & Geophysics
Li Zhu, Gongjian Wang, Faming Huang, Yan Li, Wei Chen, Haoyuan Hong
Summary: This letter proposes a sparse feature extraction network (SFE+) for landslide susceptibility prediction (LSP), which improves the accuracy of traditional machine learning models in capturing nonlinear correlations among environmental factors. The SFE-based ML models show promising prospects for LSP.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geosciences, Multidisciplinary
Faming Huang, Jun Yan, Xuanmei Fan, Chi Yao, Jinsong Huang, Wei Chen, Haoyuan Hong
Summary: In landslide susceptibility mapping (LSM), the expression of landslide boundaries and spatial shapes as points or circles instead of accurate polygons can lead to differences in the predicted landslide susceptibility indexes (LSIs) and introduce uncertainties into the LSM. This study compared the uncertainties of LSM modeling using different representations of landslide boundaries and spatial shapes, and found that using polygonal surfaces to represent the landslide boundaries can significantly improve the accuracy of LSM compared to using points and circles. The results also showed that polygon-based models have higher LSM accuracy compared to point- and circle-based models, and the overall accuracy of the random forest (RF) model is superior to that of the support vector machine (SVM) model.
GEOSCIENCE FRONTIERS
(2022)
Article
Geography, Physical
Faming Huang, Jiawu Chen, Weiping Liu, Jinsong Huang, Haoyuan Hong, Wei Chen
Summary: This study focuses on the rainfall-induced landslide hazard, using machine learning models to predict landslide susceptibility and proposing different critical rainfall threshold methods. The coupling of susceptibility maps and critical rainfall threshold values effectively predicts the rainfall-induced landslide hazards.
Article
Environmental Sciences
Loanna Ilia, Paraskevas Tsangaratos, Ploutarchos Tzampoglou, Wei Chen, Haoyuan Hong
Summary: This study introduced a novel methodological approach for flash flood susceptibility modeling based on a stacking ensemble (SE) model. The SE-RF model produced slightly higher predictive results compared to the SE-ANN model, indicating that SE models can achieve higher accuracy by intelligently combining multiple weak predictive models.
GEOCARTO INTERNATIONAL
(2022)
Article
Environmental Sciences
Yuting Li, Haoyuan Hong
Summary: This study proposes three deep learning coupling with ensemble learning models and explores the effect of these coupling methods in flood susceptibility modelling. The results show that these coupling models have reliable and excellent performance.
JOURNAL OF ENVIRONMENTAL MANAGEMENT
(2023)
Article
Biodiversity Conservation
Haoyuan Hong
Summary: Landslide susceptibility mapping is important for reducing the impact of landslides. This study proposes a hybrid model that combines the Best-first decision tree (BFT) model with other models to assess their performance. A landslide inventory map was created using 364 landslides and non-landslide data in Yongxin County, China. The Support vector machines (SVM) method was used to determine the most useful factors for modeling. The results demonstrate that the hybrid models outperform the single BFT model, with BFT-D and BFT-B being the most effective models for landslide susceptibility modeling. These models can aid in land use planning and infrastructure development in Yongxin County.
ECOLOGICAL INDICATORS
(2023)
Article
Engineering, Environmental
Haoyuan Hong
Summary: Modeling landslide susceptibility is an important technology to prevent losses from landslide disasters. This study designed three integrated models combining the multilayer perceptron model with meta classifiers to improve the accuracy of susceptibility modeling. The results show that these integrated models outperformed the MLP model and the MultiboostAB-MLP model was the most stable and effective.
BULLETIN OF ENGINEERING GEOLOGY AND THE ENVIRONMENT
(2023)
Article
Forestry
Haoyuan Hong
Summary: The aim of this paper is to design combination models to analyze the results of landslide susceptibility modeling. By combining methods such as Hoeffding tree and forest, the combination models performed well in the case study conducted in Yanshan County, Jiangxi Province, China. Among the combination models, the FPA-RF model exhibited the most stable and accurate performance.
FOREST ECOLOGY AND MANAGEMENT
(2023)
Article
Environmental Sciences
Paraskevas Tsangaratos, Ioanna Ilia, Aikaterini-Alexandra Chrysafi, Ioannis Matiatos, Wei Chen, Haoyuan Hong
Summary: The main goal of this study is to evaluate the predictive accuracy of a one-dimensional convolutional neural network model (1D-CNN) in flood susceptibility assessment in a selected test site on the island of Euboea, Greece. Different benchmark models such as logistic regression (LR), Naive Bayes (NB), gradient boosting (GB), and a deep learning neural network (DLNN) were used for performance comparison with the 1D-CNN model. Remote sensing techniques were employed to gather flood-related data, and thirteen variables were considered as predictors for flood susceptibility, including elevation, slope, curvature, topographic wetness index, lithology, sediment content, distance to faults, and distance to river network. The evaluation process involved estimating the predictive ability of all models using classification accuracy, sensitivity, specificity, and the area under the success and predictive rate curves (AUC). The analysis results showed that the 1D-CNN model achieved the highest accuracy (0.924), followed by LR (0.904) and DLNN (0.899). Overall, 1D-CNNs proved to be valuable tools for accurately analyzing flood susceptibility using remote sensing data.
Article
Computer Science, Artificial Intelligence
Haoyuan Hong
Summary: The aim of this study is to propose five integration models for assessing landslide susceptibility. By integrating different algorithms and models, it is found that the LWL-RS-ADT model is the most reliable and stable. The study also identifies NDVI, lithology, and altitude as key factors in predicting landslide susceptibility.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)