Landslide susceptibility assessment is vital for landslide risk management and urban planning, and the scientific community is continuously proposing new approaches to map landslide susceptibility, especially by hybridizing state-of-the-art models and by proposing new ones. A common practice in landslide susceptibility studies is to compare (two or more) different models in terms of AUC (area under ROC curve) to assess which one has the best predictive performance. The objective of this paper is to show that the classical scheme of comparison between susceptibility models can be expanded and enriched with substantial geomorphological insights by focusing the comparison on the mapped susceptibility values and investigating the geomorphological reasons of the differences encountered. To this aim, we used four susceptibility maps of the Wanzhou County (China) obtained with four different classification methods (namely, random forest, index of entropy, frequency ratio, and certainty factor). A quantitative comparison of the susceptibility values was carried out on a pixel-by-pixel basis, to reveal systematic spatial patterns in the differences among susceptibility maps; then, those patterns were put in relation with all the explanatory variables used in the susceptibility assessments. The lithological and morphological features of the study area that are typically associated to underestimations and overestimations of susceptibility were identified. The results shed a new light on the susceptibility models, identifying systematic errors that could be probably associated either to shortcomings of the models or to distinctive morphological features of the test site, such as nearly flat low altitude areas near the main rivers, and some lithological units.

This work presents the ideal combination of space-borne and ground-based (GB) Interferometric Synthetic Aperture Radar (InSAR) applications. In the absence of early investigation reporting and specialized monitoring, the Zhongbao landslide unexpectedly occurred on 25 July 2020, forming a barrier lake that caused an emergency. As an emergency measure, the GB-InSAR system was installed 1.8 km opposite the landslide to assess real-time cumulative deformation with a monitoring frequency of 3 min. A zone of strong deformation was detected, with 178 mm deformation accumulated within 15 h, and then a successful emergency warning was issued to evacuate on-site personnel. Post-event InSAR analysis of 19 images acquired by the ESA Sentinel-1 from December 2019 to August 2020 revealed that the landslide started in March 2020. However, the deformation time series obtained from satellite InSAR did not show any signs that the landslide had occurred. The results suggest that satellite InSAR is effective for mapping unstable areas but is not qualified for rapid landslide monitoring and timely warning. The GB-InSAR system performs well in monitoring and providing early warning, even with dense vegetation on the landslide. The results show the shortcomings of satellite InSAR and GB-InSAR and a clearer understanding of the necessity of combining multiple monitoring methods.

It is crucial to predict landslide displacement accurately for establishing a reliable early warning system. Such a requirement is more urgent for landslides in the reservoir area. The main reason is that an inaccurate prediction can lead to riverine disasters and secondary surge disasters. Machine learning (ML) methods have been developed and commonly applied in landslide displacement prediction because of their powerful nonlinear processing ability. Recently, deep ML methods have become popular, as they can deal with more complicated problems than conventional ML methods. However, it is usually not easy to obtain a well-trained deep ML model, as many hyperparameters need to be trained. In this paper, a deep ML method—the gated recurrent unit (GRU)—with the advantages of a powerful prediction ability and fewer hyperparameters, was applied to forecast landslide displacement in the dam reservoir. The accumulated displacement was firstly decomposed into a trend term, a periodic term, and a stochastic term by complementary ensemble empirical mode decomposition (CEEMD). A univariate GRU model and a multivariable GRU model were employed to forecast trend and stochastic displacements, respectively. A multivariable GRU model was applied to predict periodic displacement, and another two popular ML methods—long short-term memory neural networks (LSTM) and random forest (RF)—were used for comparison. Precipitation, reservoir level, and previous displacement were considered to be candidate-triggering factors for inputs of the models. The Baijiabao landslide, located in the Three Gorges Reservoir Area (TGRA), was taken as a case study to test the prediction ability of the model. The results demonstrated that the GRU algorithm provided the most encouraging results. Such a satisfactory prediction accuracy of the GRU algorithm depends on its ability to fully use the historical information while having fewer hyperparameters to train. It is concluded that the proposed model can be a valuable tool for predicting the displacements of landslides in the TGRA and other dam reservoirs.