Physically Based Susceptibility Assessment of Rainfall-Induced Shallow Landslides Using a Fuzzy Point Estimate Method

Displacement Characterization and Spatial-Temporal Evolution of the 2020 Aniangzhai Landslide in Danba County Using Time-Series InSAR and Multi-Temporal Optical Dataset

On 17 June 2020, a large ancient landslide over the Aniangzhai (ANZ) slope, Danba County, Sichuan Province, China, was reactivated by a series of multiple phenomena, including debris flow triggered by heavy rainfall and flooding. In this study, Synthetic Aperture Radar (SAR) images acquired by the Sentinel-1A/B satellite and optical images captured by the PlanetScope satellites were jointly used to analyze and explore the deformation characteristics and the Spatial-Temporal evolution of the ANZ landslide before and after the multi-hazard chain. Several areas of pre-failure movements were found from the multi-temporal optical images analysis before the reactivation of the ANZ landslide. The large post-failure surface deformation over the ANZ slope was also retrieved by the optical pixel offset tracking (POT) technique. A major northwest movement with the maximum horizontal deformation of up to 14.4 m was found. A time-series InSAR technique was applied to analyze the descending and ascending Sentinel-1A/B datasets spanning from March 2018 to July 2020, showing that the maximum magnitudes of the Line of Sight (LoS) displacement velocities were −70 mm/year and 45 mm/year, respectively. The Spatial-Temporal evolution over the ANZ landslide was analyzed based on the time-series results. No obvious change in acceleration (precursory deformation) was detected before the multi-hazard chain, while clear accelerated deformation can be observed over the slope after the event. This suggested that heavy rainfall was the most significant triggering factor for the generation and reactivation of the ANZ landslide. Other preparatory factors, including the deformation behavior, the undercutting and erosion of the river and the outburst flood, the local terrain conditions, and earthquakes, might also have played an important role in the generation and reactivation of the landslide.

Regional Analyses of Rainfall-Induced Landslide Initiation in Upper Gudbrandsdalen (South-Eastern Norway) Using TRIGRS Model

In Norway, shallow landslides are generally triggered by intense rainfall and/or snowmelt events. However, the interaction of hydrometeorological processes (e.g., precipitation and snowmelt) acting at different time scales, and the local variations of the terrain conditions (e.g., thickness of the surficial cover) are complex and often unknown. With the aim of better defining the triggering conditions of shallow landslides at a regional scale we used the physically based model TRIGRS (Transient Rainfall Infiltration and Grid-based Regional Slope stability) in an area located in upper Gudbrandsdalen valley in South-Eastern Norway. We performed numerical simulations to reconstruct two scenarios that triggered many landslides in the study area on 10 June 2011 and 22 May 2013. A large part of the work was dedicated to the parameterization of the numerical model. The initial soil-hydraulic conditions and the spatial variation of the surficial cover thickness have been evaluated applying different methods. To fully evaluate the accuracy of the model, ROC (Receiver Operating Characteristic) curves have been obtained comparing the safety factor maps with the source areas in the two periods of analysis. The results of the numerical simulations show the high susceptibility of the study area to the occurrence of shallow landslides and emphasize the importance of a proper model calibration for improving the reliability.

Landslide Susceptibility Prediction Based on Remote Sensing Images and GIS: Comparisons of Supervised and Unsupervised Machine Learning Models

Landslide susceptibility prediction (LSP) has been widely and effectively implemented by machine learning (ML) models based on remote sensing (RS) images and Geographic Information System (GIS). However, comparisons of the applications of ML models for LSP from the perspectives of supervised machine learning (SML) and unsupervised machine learning (USML) have not been explored. Hence, this study aims to compare the LSP performance of these SML and USML models, thus further to explore the advantages and disadvantages of these ML models and to realize a more accurate and reliable LSP result. Two representative SML models (support vector machine (SVM) and CHi-squared Automatic Interaction Detection (CHAID)) and two representative USML models (K-means and Kohonen models) are respectively used to scientifically predict the landslide susceptibility indexes, and then these prediction results are discussed. Ningdu County with 446 recorded landslides obtained through field investigations is introduced as case study. A total of 12 conditioning factors are obtained through procession of Landsat TM 8 images and high-resolution aerial images, topographical and hydrological spatial analysis of Digital Elevation Modeling in GIS software, and government reports. The area value under the curve of receiver operating features (AUC) is applied for evaluating the prediction accuracy of SML models, and the frequency ratio (FR) accuracy is then introduced to compare the remarkable prediction performance differences between SML and USML models. Overall, the receiver operation curve (ROC) results show that the AUC of the SVM is 0.892 and is slightly greater than the AUC of the CHAID model (0.872). The FR accuracy results show that the SVM model has the highest accuracy for LSP (77.80%), followed by the CHAID model (74.50%), the Kohonen model (72.8%) and the K-means model (69.7%), which indicates that the SML models can reach considerably better prediction capability than the USML models. It can be concluded that selecting recorded landslides as prior knowledge to train and test the LSP models is the key reason for the higher prediction accuracy of the SML models, while the lack of a priori knowledge and target guidance is an important reason for the low LSP accuracy of the USML models. Nevertheless, the USML models can also be used to implement LSP due to their advantages of efficient modeling processes, dimensionality reduction and strong scalability.