Artificial neural networks applied to landslide susceptibility: The effect of sampling areas on model capacity for generalization and extrapolation

Gameiro, Samuel; Riffel, Eduardo Samuel; Oliveira, Guilherme Garcia de; Guasselli, Laurindo Antônio

doi:10.1016/j.apgeog.2021.102598

Cited by 28 publications

(4 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Masruroh et al (2023) used ANN to evaluate the susceptibility to landslides, their findings revealed that the AUC of the model reached 96.5%, which is comparable to our study using ANN, in which the AUC is approximately 88%, and for the ensemble of ANN and LR, the AUC approaches 91%. In addition, our study also is compatible with the study of Gameiro et al (2021), where they used ANN, and the AUC for prediction reached 93.1%.…”

Section: Discussionsupporting

confidence: 79%

See 1 more Smart Citation

Landslide susceptibility, ensemble machine learning, and accuracy methods in the southern Sinai Peninsula, Egypt: Assessment and Mapping

Youssef,

El‑Haddad,

Skilodimou

et al. 2023

Preprint

View full text Add to dashboard Cite

Each year, thousands of tourists visit Egypt's Wadi Feiran region, one of the most popular tourist sites in the Sinai Peninsula. The region’s topography is distinctive and diverse, making it particularly susceptible to “natural disasters” (such as floods and landslides). The current study deals with landslide hazards as a critical hazard type, where, after rainfall, hundreds of landslides occur annually, and landslide disaster assessments are becoming more necessary to reduce mountain hazards. The current research mapped "landslide susceptibility" in the Wadi Feiran basin using three different modeling strategies: “Logistic Regression” -LR, “Artificial Neural Network”-ANN, and an "ensemble" of LR and ANN. A “landslides” map was first created as a preliminary stage, using 800 landslide locations acquired from multiple data sources (30% validation datasets, 70% training datasets), including historical records, field surveys, and high-resolution satellite imagery. In addition, fourteen landslide causative parameters (LCPs), elevation (El), “distance to wadis” (DtW), “distance to fault” (DtF), “distance to road” (DtR), lithology (Lith), aspect (As), “profile-curvature” (PrC), “plan-curvature” (PlC), “slope length” (LS), slope-angle (Sa), “topographic wetness index” (TWI), “relative slope position” (RSP), rainfall (Ra), and “topographic roughness index” (TRI) were employed. These models' accuracy was evaluated using “receiver operating characteristics and area under the curve (ROC - AUC),” “root mean square error”-RMSE, and “kappa index”-K. According to the findings, the AUC for LR, ANN, and ensemble of LR &ANN were 82%, 89%, and 91%, respectively. The results showed that the ensemble model outperformed ANN and LR by 2.3% and 10.9%, respectively, whereas ANN model outperformed LR by 8.5%. Other statistical indices also revealed that the RMSE and kappa index values obtained by LR were the highest and the lowest, respectively, whereas the RMSE and kappa index values generated by the LR&ANN ensemble were the lowest and the highest, respectively. These results indicate that landslides are influenced by a wide variety of natural and anthropogenic factors. To better manage and avoid landslides, it is important to create maps that show the area’s most susceptible to landslides.

show abstract

Section: Discussionsupporting

confidence: 79%

“…ANN has the advantages of high modeling efficiency, simple application, and high stability. Data entering this complex network of neurons is processed based on the connections between neurons, continuously learning the relationship between the input and output by changing the weight values (Gameiro et al, 2021).…”

Section: Annmentioning

confidence: 99%

Landslide susceptibility, ensemble machine learning, and accuracy methods in the southern Sinai Peninsula, Egypt: Assessment and Mapping

Youssef,

El‑Haddad,

Skilodimou

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Many researchers have devised efficient methods for creating an accurate landslide susceptibility map over the last few decades. Frequency ratio (Goetz et al, 2015;Hong et al, 2016;Lee et al, 2016), logistic regression (Chen et al, 2017;Steger et al, 2016), decision trees (Beucher et al, 2019), fuzzy logic (Pham et al, 2021), neuro-fuzzy systems (Shihabudheen and Pillai, 2018), support vector machines (Huang and Zhao, 2018), artificial neural networks (Gameiro et al, 2021), Analytical Hierarchy Process (AHP) (Sonker et al, 2021), information value method (IVM) (Farooq and Akram, 2021) and multimethod approach (Wubalem, 2021) are a few illustrations of these methods. The methodology used a frequency ratio (FR) to evaluate the efficiency of the landslide susceptibility study.…”

Section: Methodsmentioning

confidence: 99%

Landslide susceptibility mapping of the Main Boundary Thrust (MBT) region in Tinau-Mathagadhi Section of Palpa District, Lumbini Province

Paudyal

Maharjan

2022

Journal of Nepal Geological Society

View full text Add to dashboard Cite

The study was carried out in Tinau-Mathagadhi section of Palpa District, Lumbini Province. The output of the landslide susceptibility analysis using frequency ratio (FR) is evaluated. Google Earth (CNES/Airbus and Maxar Technologies) Imagery of 50 cm spatial resolution was used to detect landslides. Data training and testing sampling was created using the landslide inventory. Eight causative factors were derived from topographic, geological, and land-use maps. The causative factors and training events were used to determine the FR ratings. The objective of present study is to address the effect of the Main Boundary Thrust (MBT) along with other important parameters in landslide susceptibility mapping. The integration of causative factors that assigned FR scores yielded the landslide susceptibility map. The validation of 79.6% was obtained using the ROC-AUC curve. Among eight causative factors, it was found that distance from the Main Boundary Thrust (MBT), aspect, land-use, and geology are dominating factors in the occurrence of the landslide. The FR method is based on a quantitative relationship between landslide inventory and landslide affecting factors. The method is valid for Tinau-Mathagadhi section and MBT is one of the causes of landslides in this area.

show abstract

“…Statistical LSA is a supervised dichotomy problem, which can be solved via different classification methods [7]. About 163 different data-driven methods are applied to LSA [8], such as weight of evidence (WoE) [9,10], naïve Bayes (NB) [6,11], logistic regression (LR) [12][13][14][15], discriminant analysis (DA) [3,16], supported vector machines (SVM) [17,18], random forest (RF) [13,[19][20][21], artificial neural networks (ANN) [22][23][24], and many others. These methods have their advantages and disadvantages.…”

Section: Introductionmentioning

confidence: 99%

Modeling and Assessment of Landslide Susceptibility of Dianchi Lake Watershed in Yunnan Plateau

Bai,

Yang,

Kong

et al. 2023

Sustainability

View full text Add to dashboard Cite

The nine plateau lake watersheds in Yunnan are important ecological security barriers in the southwest of China. The prevention and control of landslides are important considerations in the management of these watersheds. Taking the Dianchi Lake watershed as a typical research area, a comprehensive modeling and assessment process of landslide susceptibility was put forward. The comprehensive process was based on the weight of evidence (WoE) method, and many statistical techniques were integrated, such as cross-validation, multi-quantile cumulative Student’s comprehensive weight statistics, independence testing, step-by-step modeling, ROC analysis, and ROC-based susceptibility zoning. In this paper, fourteen models with high accuracy and validity were established, and the AUC reached 0.83–0.87 and 0.85–0.88, respectively. In addition, according to the susceptibility zoning map compiled via the optimal model, 80% of landslides can be predicted in the very-high- and high-susceptibility areas, which only account for 19.58% of the study area. Finally, this paper puts forward strategies for geological disaster prevention and ecological restoration deployment.

show abstract

Artificial neural networks applied to landslide susceptibility: The effect of sampling areas on model capacity for generalization and extrapolation

Cited by 28 publications

References 54 publications

Landslide susceptibility, ensemble machine learning, and accuracy methods in the southern Sinai Peninsula, Egypt: Assessment and Mapping

Landslide susceptibility, ensemble machine learning, and accuracy methods in the southern Sinai Peninsula, Egypt: Assessment and Mapping

Landslide susceptibility mapping of the Main Boundary Thrust (MBT) region in Tinau-Mathagadhi Section of Palpa District, Lumbini Province

Modeling and Assessment of Landslide Susceptibility of Dianchi Lake Watershed in Yunnan Plateau

Contact Info

Product

Resources

About