1.Uncertainties of prediction accuracy in shallow landslide modeling: Sample size and raster resolution
Shirzadi, Ataollah; Solaimani, Karim; Roshan, Mahmood Habibnejad;
CATENA, Volume 178, Pages 172-188, JUL 2019
[Abstract]Understanding landslide characteristics such as their locations, dimensions, and spatial distribution is of highly importance in landslide modeling and prediction. The main objective of this study was to assess the effect of different sample sizes and raster resolutions in landslide susceptibility modeling and prediction accuracy of shallow landslides. In this regard, the Bijar region of the Kurdistan province (Iran) was selected as a case study. Accordingly, a total of 20 landslide conditioning factors were considered with six different raster resolutions (10 m, 15 m, 20 m, 30 m, 50 m, and 100 m) and four different sample sizes (60/40%, 70/30%, 80/20%, and 90/10%) were investigated. The merit of each conditioning factors was assessed using the Information Gain Ratio (IGR) technique, whereas Alternating decision tree (ADTree), which has been rarely explored for landslide modeling, was used for building models. Performance of the models was assessed using the area under the ROC curve (AUROC), sensitivity, specificity, accuracy, kappa and RMSE criteria. The results show that with increasing the number of training pixels in the modeling process, the accuracy is increased. Findings also indicate that for the sample sizes of 60/40% (AUROC = 0.800) and 70/30% (AUROC = 0.899), the highest prediction accuracy is derived with the raster resolution of 10 m. With the raster resolution of 20 m, the highest prediction accuracy for the sample size of 80/20% (AUROC = 0.871) and 90/10% (AUROC = 0.864). These outcomes provide a guideline for future research enabling researchers to select an optimal data resolution for landslide hazard modeling.
2.Spatial prediction of landslide susceptibility by combining evidential belief function, logistic regression and logistic model tree
Chen, Wei; Zhao, Xia; Shahabi, Himan;
GEOCARTO INTERNATIONAL, Volume 34, Issue 11 , Pages 1177-1201, SEP 19 2019
[Abstract]In this study, we introduced novel hybrid of evidence believe function (EBF) with logistic regression (EBF-LR) and logistic model tree (EBF-LMT) for landslide susceptibility modelling. Fourteen conditioning factors were selected, including slope aspect, elevation, slope angle, profile curvature, plan curvature, topographic wetness index (TWI), stream sediment transport index (STI), stream power index (SPI), distance to rivers, distance to faults, distance to roads, lithology, normalized difference vegetation index (NDVI), and land use. The importance of factors was assessed using correlation attribute evaluation method. Finally, the performance of three models was evaluated using the area under the curve (AUC). The validation process indicated that the EBF-LMT model acquired the highest AUC for the training (84.7%) and validation (76.5%) datasets, followed by EBF-LR and EBF models. Our result also confirmed that combination of a decision tree-logistic regression-based algorithm with a bivariate statistical model lead to enhance the prediction power of individual landslide models.
3.Landslide spatial modelling using novel bivariate statistical based Naive Bayes, RBF Classifier, and RBF Network machine learning algorithms
He, Qingfeng; Shahabi, Himan; Shirzadi, Ataollah;
SCIENCE OF THE TOTAL ENVIRONMENT, Volume 663 , Pages 1-15, MAY 1 2019
[Abstract]Landslides are major hazards for human activities often causing great damage to human lives and infrastructure. Therefore, the main aim of the present study is to evaluate and compare three machine learning algorithms (MLAs) including Naive Bayes (NB), radial basis function (RBF) Classifier, and RBF Network for landslide susceptibilitymapping (LSM) at Longhai area in China. A total of 14 landslide conditioning factors were obtained from various data sources, then the frequency ratio (FR) and support vector machine (SVM) methods were used for the correlation and selection the most important factors for modelling process, respectively. Subsequently, the resulting three models were validated and compared using some statistical metrics including area under the receiver operating characteristics (AUROC) curve, and Friedman and Wilcoxon signed-rank tests The results indicated that the RBF Classifier model had the highest goodness-of-fit and performance based on the training and validation datasets. The results concluded that the RBF Classifier model outperformed and outclassed (AUROC = 0.881), the NB (AUROC = 0.872) and the RBF Network (AUROC = 0.854) models. The obtained results pointed out that the RBF Classifier model is a promising method for spatial prediction of landslide over the world. (C) 2019 Published by Elsevier B.V.
4.Landslide susceptibility assessment at the Wuning area, China: a comparison between multi-criteria decision making, bivariate statistical and machine learning methods
Hong, Haoyuan; Shahabi, Himan; Shirzadi, Ataollah;
NATURAL HAZARDS, Volume 96, Issue 1, Pages 173-212, MAR 2019
[Abstract]The aim of this research is to investigate multi-criteria decision making [spatial multi-criteria evaluation (SMCE)], bivariate statistical methods [frequency ratio (FR), index of entropy (IOE), weighted linear combination (WLC)] and machine learning [support vector machine (SVM)] models for estimating landslide susceptibility at the Wuning area, China. A total of 445 landslides were randomly classified into 70% (311 landslides) and 30% (134 landslides) to train and validate landslide models, respectively. Fourteen landslide conditioning factors including slope angle, slope aspect, altitude, topographic wetness index, stream power index, sediment transport index, soil, lithology, NDVI, land use, rainfall, distance to road, distance to river and distance to fault were then studied for landslidesusceptibility assessment. Performances of five studied models were evaluated using area under the ROC curve (AUROC) for training (success rate curve) and validation (prediction rate curve) datasets, statistical-based measures and tests. Results indicated that the area under the success rate curve for the FR, IOE, WLC, SVM and SMCE models was 88.32%, 82.58%, 78.91%, 85.47% and 89.96%, respectively, demonstrating that SMCE could provide the higher accuracy. The prediction capability findings revealed that the SMCE model (AUC=86.81%) was also the highest approach among the five studied models, followed by the FR (AUC=84.53%), the SVM (AUC=81.24%), the IOE (AUC=79.67%) and WLC (73.92%) methods. The landslide susceptibility maps derived from the above five models are reasonably accurate and could be used to perform elementary land use planning for hazard extenuation.
5.Hybrid Machine Learning Approaches for Landslide Susceptibility Modeling
Vu Viet Nguyen; Binh Thai Pham; Ba Thao Vu;
FORESTS, Volume 10 , Issue2 , Article 157, FEB 2019
[Abstract]This paper presents novel hybrid machine learning models, namely Adaptive Neuro Fuzzy Inference System optimized by Particle Swarm Optimization (PSOANFIS), Artificial Neural Networks optimized by Particle Swarm Optimization (PSOANN), and Best First Decision Trees based Rotation Forest (RFBFDT), for landslide spatial prediction. Landslide modeling of the study area of Van Chan district, Yen Bai province (Vietnam) was carried out with the help of a spatial database of the area, considering past landslides and 12 landslide conditioning factors. The proposed models were validated using different methods such as Area under the Receiver Operating Characteristics (ROC) curve (AUC), Mean Square Error (MSE), Root Mean Square Error (RMSE). Results indicate that the RFBFDT (AUC = 0.826, MSE = 0.189, and RMSE = 0.434) is the best method in comparison to other hybrid models, namely PSOANFIS (AUC = 0.76, MSE = 0.225, and RMSE = 0.474) and PSOANN (AUC = 0.72, MSE = 0.312, and RMSE = 0.558). Thus, it is reasonably concluded that the RFBFDT is a promising hybrid machine learning approach for landslide susceptibilitymodeling.
6.Applying population-based evolutionary algorithms and a neuro-fuzzy system for modeling landslidesusceptibility
Chen, Wei; Panahi, Mandi; Tsangaratos, Paraskevas;
CATENA, Volume 172 , Pages 212-231, JAN 2019
[Abstract]The main objective of the present study was to produce a novel ensemble data mining technique that involves an adaptive neuro-fuzzy inference system (ANFIS) optimized by Shuffled Frog Leaping Algorithm (SFLA) and Particle Swarm Optimization (PSO) for spatial modeling of landslide susceptibility. Step-wise Assessment Ratio Analysis (SWARA) was utilized for the evaluation of the relation between landslides and landslide-related factors providing ANFIS with the necessary weighting values. The developed methods were applied in Langao County, Shaanxi Province, China. Eighteen factors were selected based on the experience gained from studying landslidephenomena, the local geo-environmental conditions as well as the availability of data, namely; elevation, slope aspect, slope angle, profile curvature, plan curvature, sediment transport index, stream power index, topographic wetness index, land use, normalized difference vegetation index, rainfall, lithology, distance to faults, fault density, distance to roads, road density, distance to rivers and river density. A total of 288 landslides were identified after analyzing previous technical surveys, airborne imagery and conducting field surveys. Also, 288 non-landslide areas were identified with the usage of Google Earth imagery and the analysis of a digital elevation model. The two datasets were merged and later divided into two subsets, training and testing, based on a random selection scheme. The produced landslide susceptibility maps were evaluated by the receiving operating characteristic and the area under the success and predictive rate curves (AUC). The results showed that AUC based on the training and testing dataset was similar and equal to 0.89. However, the processing time during the training and implementation phase was considerable different. SWARA-ANFIS-PSO appeared six times faster in respect to the processing time achieved by SWARA-ANFIS-SFLA. The proposed novel approach, which combines expert knowledge, neuro-fuzzy inference systems and evolutionary algorithms, can be applied for land use planning and spatial modeling of landslidesusceptibility.
7.A review of statistically-based landslide susceptibility models
Reichenbach, Paola; Rossi, Mauro; Malamud, Bruce D.;
EARTH-SCIENCE REVIEWS, Volume 180, Pages 60-91, MAY 2018
[Abstract]In this paper, we do a critical review of statistical methods for landslide susceptibility modelling and associated terrain zonations. Landslide susceptibility is the likelihood of a landslide occurring in an area depending on local terrain conditions, estimating "where" landslides are likely to occur. Since the first attempts to assess landslidesusceptibility in the mid-1970s, hundreds of papers have been published using a variety of approaches and methods in different geological and climatic settings. Here, we critically review the statistically-based landslide susceptibilityassessment literature by systematically searching for and then compiling an extensive database of 565 peer-review articles from 1983 to 2016. For each article in the literature database, we noted 31 categories/sub-categories of information including study region/extent, landslide type/number, inventory type and period covered, statistical model used, including variable types, model fit/prediction performance evaluation method, and strategy used to assess the model uncertainty. We present graphical visualisations and discussions of commonalities and differences found as a function of region and time, revealing a significant heterogeneity of thematic data types and scales, modelling approaches, and model evaluation criteria. We found that the range of thematic data types used for susceptibility assessment has not changed significantly with time, and that for a number of studies the geomorphological significance of the thematic data used is poorly justified. We also found that the most common statistical methods for landslide susceptibility modelling include logistic regression, neural network analysis, data-overlay, index-based and weight of evidence analyses, with an increasing preference towards machine learning methods in the recent years. Although an increasing number of studies in recent years have assessed the model performance, in terms of model fit and prediction performance, only a handful of studies have evaluated the model uncertainty. Adopting a Susceptibility Quality Level index, we found that the quality of published models has improved over the years, but top-quality assessments remain rare. We identified a clear geographical bias in susceptibility study locations, with many studies in China, India, Italy and Turkey, and only a few in Africa, South America and Oceania. Based on previous literature reviews, the analysis of the information collected in the literature database, and our own experience on the subject, we provide recommendations for the preparation, evaluation, and use of landslide susceptibility models and associated terrain zonations.