Search

Machine learning models have been widely used in landslide susceptibility prediction. However, landslide multidimensional feature extraction, model generalization ability, and prediction quantification of deep learning are still challenging. This paper proposes a new approach, the landslide density-based bidirectional long short-term memory (LD-BiLSTM) model with multichannel input and an optimized sampling strategy to predict and map landslide susceptibility in active seismic mountainous areas of Sichuan Province, China. First, to ensure the generalization ability of the LD-BiLSTM model, other regions in Sichuan were selected as the model training area independent of the prediction area (Luding County). Multichannel landslide datasets were constructed to extract high-dimensional geospatial features of landslides. Subsequently, the landslide density of each grid cell was utilized as the label for the corresponding input sample. The LD-BiLSTM model was improved by using transfer learning and sampling optimization strategies, which makes our method attenuate the impact of historical landslide inventory deviation on the spatial susceptibility mode compared with the existing DL model, which usually uses landslide objects (LO) as input sample labels. Model performance evaluation results show that the LD-BiLSTM model (precision = 0.903, recall = 0.899, F1-score = 0.901, Area under receiver operating characteristic curve (AUC) = 0.940) outperformed the LO-BiLSTM model (precision = 0.812, recall = 0.815. F1-score = 0.813, AUC = 0.910) in the case areas. Meanwhile, the performance of the LD-BiLSTM model (AUC = 0.9407) significantly outperformed both the information value (IV) (AUC = 0.7207) model and the random forest (RF) (AUC = 0.8116) models in the landslide prediction area (Luding), which confirms that the proposed LD-based method is superior to traditional LO-based methods. Significantly, our approach can effectively extract the spatial distribution of landslides and predict potential landslides in complex high-mountain environments. [ABSTRACT FROM AUTHOR]

Copyright of Revista de la Asociación Geológica Argentina is the property of Asociacion Geologica Argentina and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

This paper presents a new method to estimate a landslide failure surface and its volume based on a digital elevation model (DEM) using spline curves by assuming values of the tangent of the failure surface along with transversal vertical profiles. The model will give the depth of the probable failure surface plotted using the 2D grid function. We can easily visualize the results by running the codes; the display will show the step-by-step processes involved. The model requires fundamental data inputs, which are readily available online throughout the world, such as a DEM of the study area and a KML file of contour limits of the landslide as input. The calculation procedure is simple; it evaluates the depth of the failure surface at each point of the grid. The cross-section for each point is estimated perpendicular to the slope line joining the highest elevation point and the lowest elevation point within the contour limits. A cubic spline curve is calculated using two endpoints and the first derivative at these points. The Z value is calculated using these parameters at each grid point (centre of a grid), and finally, a 3D failure surface is generated, enabling the volume calculation. MATLAB code has been written to execute this process automatically. This model can easily use input data with very little fieldwork, and the processing can produce a result in a few minutes. We can enrich a real scenario by imposing more constraints on the variation in the cross-section angle based on the limit of the landslide using field measurements. This model is also helpful in predicting the area affected by the mass of the failure surface, and we can determine a predefined hazard zone to avoid any casualties in the future. [ABSTRACT FROM AUTHOR]

The author's models revealed that the detection of landslide sources provided accurate results, while the extraction of the runout areas achieved excellent accuracies. Lai [[11]] performed an automated data-mining procedure to differentiate the landslide sources and runout zones of landslides triggered by Typhoon Morakot in Taiwan. The authors used ground-based InSAR to monitor slope deformations and verified the method on five landslides in the area of Huangdao, China. Keywords: landslide; monitoring; modelling; susceptibility; InSAR EN landslide monitoring modelling susceptibility InSAR 10453 1 11/17/21 20211101 NES 211101 1. [Extracted from the article]

The generation of a tsunami by a landslide is a complex phenomenon that involves landslide dynamics, wave dynamics and their interaction. Numerous lives and infrastructures around the world are threatened by this phenomenon. Predictive numerical models are a suitable tool to assess this natural hazard. However, the complexity of this phenomenon causes such models to be either computationally inefficient or unable to handle the overall process. Our model, which is based on shallow-water equations, has been developed to address these two problems. In our model, the two materials are treated as two different layers, and their interaction is resolved by momentum transfer inspired by elastic collision principles. The goal of this study is to demonstrate the validity of our model through benchmark tests based on physical experiments performed by Miller et al. (2017). A dry case is reproduced to validate the behaviour of the landslide propagation model using different rheological laws and to determine which law performs best. In addition, a wet case is reproduced to investigate the influence of different still-water levels on both the landslide deposit and the generated waves. The numerical results are in good agreement with the physical experiments, thereby confirming the validity of our model, particularly concerning the novel momentum transfer approach. [ABSTRACT FROM AUTHOR]

Superelevation is an often observed phenomenon in landslide and debris flow down a complex three-dimensional topography. The degree of superelevation is controlled by the geometry of the channel, the material involved and also the flow dynamics. Empirical methods are usually applied to estimate superelevation. However, those models are incomplete and lack important aspects of the channel geometry, material properties and flow dynamics, instigating serious errors in estimating flow velocities that could only be controlled empirically by introducing ad hoc correction factors. Here, we present new and complete analytical models for superelevation and superelevation velocities down a general topography providing a fully dynamical method. New models formally include essential forces that play an important role in the flow dynamics, namely gravitational forces, topographic- and hydraulic-pressure gradients and Coulomb friction. We discuss the importance of geometry in inducing superelevation and that one directional channel without twist cannot produce superelevation. With the new models we can, in principle, exactly obtain the flow velocities of deformable landslide, dynamic impact pressures and the explicit description of deposition. We have formally provided two alternative analytical representations for superelevation: geometrical and dynamical definitions of superelevation, which is a new concept. We proved that for superelevation to take place, the transversal velocity must have a gradient across the channel. We have analytically constructed a new non-dimensional superelevation number. Superelevation velocity appears to be a non-linear function of the superelevation number. We can now explicitly quantify the superelevation intensity in landslide motion. It has several implications. We proved that superelevation is higher for fluid-saturated debris flows than for dry granular flows. New superelevation models have been validated against a laboratory granular flow down a multi-dimensionally curved channel, and a natural debris flow event in Chamoson, Valais, Switzerland. Our theoretical superelevation velocities appear to be very close to the velocity measured in laboratory and in the field, which however, are largely overestimated by the empirical models. We further validated the model by constructing an exact analytical solution and by applying it to describe superelevation-induced propagation and deposition of the natural debris flow event. New simulations produced observed propagating fronts. [ABSTRACT FROM AUTHOR]

A new inventory of 66 small and shallow landslides within six pilot areas was created based on a high-resolution digital elevation model in the canton of Vaud in Switzerland. The geometrical characteristics of the landslides were recorded (i.e. surface area, maximum thickness and length), and the volumes were estimated. These data permitted the development of a model that provides the probability for a landslide to possess a maximum thickness or volume smaller than a given value based on the landslide horizontal surface area. The results are compared with three existing power-law relationships of surface area–volumes. This new approach constitutes a way to improve the quantification of the uncertainty of volume and maximum depth estimations for small and shallow landslides. [ABSTRACT FROM AUTHOR]

In the morning of 23 August 2017, around 3 ×106 m 3 of granitoid rock broke off from the eastern face of Piz Cengalo, southeastern Switzerland. The initial rockslide–rockfall entrained 6 ×105 m 3 of a glacier and continued as a rock (or rock–ice) avalanche before evolving into a channelized debris flow that reached the village of Bondo at a distance of 6.5 km after a couple of minutes. Subsequent debris flow surges followed in the next hours and days. The event resulted in eight fatalities along its path and severely damaged Bondo. The most likely candidates for the water causing the transformation of the rock avalanche into a long-runout debris flow are the entrained glacier ice and water originating from the debris beneath the rock avalanche. In the present work we try to reconstruct conceptually and numerically the cascade from the initial rockslide–rockfall to the first debris flow surge and thereby consider two scenarios in terms of qualitative conceptual process models: (i) entrainment of most of the glacier ice by the frontal part of the initial rockslide–rockfall and/or injection of water from the basal sediments due to sudden rise in pore pressure, leading to a frontal debris flow, with the rear part largely remaining dry and depositing mid-valley, and (ii) most of the entrained glacier ice remaining beneath or behind the frontal rock avalanche and developing into an avalanching flow of ice and water, part of which overtops and partially entrains the rock avalanche deposit, resulting in a debris flow. Both scenarios can – with some limitations – be numerically reproduced with an enhanced version of the two-phase mass flow model (Pudasaini, 2012) implemented with the simulation software r.avaflow, based on plausible assumptions of the model parameters. However, these simulation results do not allow us to conclude on which of the two scenarios is the more likely one. Future work will be directed towards the application of a three-phase flow model (rock, ice, and fluid) including phase transitions in order to better represent the melting of glacier ice and a more appropriate consideration of deposition of debris flow material along the channel. [ABSTRACT FROM AUTHOR]

A landslide-generated tsunami is a complex phenomenon that involves landslide dynamics, wave dynamics and their interaction. Numerous lives and infrastructures around the world are threatened by these events. Predictive numerical models are a suitable tool to assess this natural hazard. However, the complexity of this phenomenon causes such models to be either computationally inefficient or unable to handle the overall process. Our model, which is based on shallow water equations, is developed to address these two problems. In our model, the two materials are treated as two different layers, and their interaction is resolved by momentum transfer inspired by elastic collision principles. The goal of this study is to demonstrate the validity of our model through benchmark tests based on physical experiments performed by Miller et al. (2017). A dry case is reproduced to validate the behaviour of the landslide propagation model using different rheological laws and to determine which law performs the best. In addition, a wet case is reproduced to investigate the influence of different still water levels on both the landslide deposit and the generated waves. The numerical results are in good agreement with the physical experiments, thereby confirming the validity of our model, particularly concerning the novel momentum transfer approach. [ABSTRACT FROM AUTHOR]

In the morning of 23 August 2017, around 3 million m³ of granitoid rock broke off from the east face of Piz Cengalo, SE Switzerland. The initial rock slide-rock fall entrained 0.6 million m³ of a glacier and continued as a rock(-ice) avalanche, before evolving into a channelized debris flow that reached the village of Bondo at a distance of 6.5 km after a couple of minutes. Subsequent debris flow surges followed in the next hours and days. The event resulted in eight fatalities along its path and severely damaged Bondo. The most likely candidates for the water causing the transformation of the rock avalanche into a long-runout debris flow are the entrained glacier ice and water originating from the debris beneath the rock avalanche. In the present work we try to reconstruct conceptually and numerically the cascade from the initial rock slide-rock fall to the first debris flow surge and thereby consider two scenarios in terms of qualitative conceptual process models: (i) entrainment of most of the glacier ice by the frontal part of the initial rock slide-rock fall and/or injection of water from the basal sediments due to sudden rise in pore pressure, leading to a frontal debris flow, with the rear part largely remaining dry and depositing mid-valley; and (ii) most of the entrained glacier ice remaining beneath/behind the frontal rock avalanche, and developing into an avalanching flow of ice and water, part of which overtops and partially entrains the rock avalanche deposit, resulting in a debris flow. Both scenarios can be numerically reproduced with the two-phase mass flow model implemented with the simulation software r.avaflow, based on plausible assumptions of the model parameters. However, these simulation results do not allow to conclude on which of the two scenarios is the more likely one. Future work will be directed towards the application of a three-phase flow model (rock, ice, fluid) including phase transitions, in order to better represent the melting of glacier ice, and a more appropriate consideration of deposition of debris flow material along the channel. [ABSTRACT FROM AUTHOR]

Reconstructions of failure surfaces (prior to potential landslides or after their release), landslide deposits, or other palaeotopographic features are important for hazard and erosion assessment. The volumes involved in landslide and failure surfaces constrain the propagation of a landslide, and knowledge of the past topography helps us to understand these hazards. Some methods exist to characterise landslide geometry, but these methods usually require monitoring information. This study tries to assess the validity of the sloping local base level (SLBL) method for this purpose. Two sets of airborne lidar digital elevation models (DEMs) of the Kii Peninsula (Japan) are used: the first one was acquired before Typhoon Talas, and the second one was acquired after. A total of 70 deep-seated landslides occurred during this event between 2 and 5 September 2011. This study shows that the SLBL method is efficient using either the slope deformations identifiable on the DEM before the release of the landslide or a reliable 2.5-D failure surface created by using both DEMs (the 2.5-D corresponds to a surface which has only one z value for each x – y coordinate; in other words, no true vertical topography or overhang can be represented perfectly). In addition, this method allows for the reconstruction of eroded deposits and buried valleys. Most of the volumes estimated are within ±35 % of the estimation made by Chigira et al. (2013), and the coefficients of expansion range from 10 % to 25 %. These results show considerable sensitivity to the parameters used for the reconstruction of the landslide volume estimations and demonstrate the need for an efficient and fast tool to reconstruct potential landslide geometries or histories. [ABSTRACT FROM AUTHOR]

We present an automated terrestrial laser scanning (ATLS) system with automatic near-real-time change detection processing. The ATLS system was tested on the Séchilienne landslide in France for a 6-week period with data collected at 30 min intervals. The purpose of developing the system was to fill the gap of high-temporal-resolution TLS monitoring studies of earth surface processes and to offer a cost-effective, light, portable alternative to ground-based interferometric synthetic aperture radar (GB-InSAR) deformation monitoring. During the study, we detected the flux of talus, displacement of the landslide and pre-failure deformation of discrete rockfall events. Additionally, we found the ATLS system to be an effective tool in monitoring landslide and rockfall processes despite missing points due to poor atmospheric conditions or rainfall. Furthermore, such a system has the potential to help us better understand a wide variety of slope processes at high levels of temporal detail. [ABSTRACT FROM AUTHOR]

Highlights from the article: The International School on "LAndslide Risk Assessment and Mitigation" (LARAM) (http://www.laram.unisa.it) was founded at the University of Salerno in 2005 by Prof. Leonardo Cascini, who served as the Director of the School from 2005 to 2017, within the geotechnical engineering research group of the Department of Civil Engineering. Since 2006, 11 editions were held in Salerno, Italy, in the month of September and three editions were held in Chengdu, China. Graph: Fig. 2Pictures from the last edition of the LARAM doctoral school, held at the University of Salerno in September 2018.

Monitoring landslides with terrestrial laser scanning (TLS) is currently a well-known technique. One problem often encountered is the vegetation that produces shadow areas on the scans. Indeed, the points behind a given obstacle are hidden and thus occluded on the point cloud. Thereby, locations monitored with terrestrial laser scanner are mostly rock instabilities and few vegetated landslides, being difficult or even impossible to survey vegetated slopes using this method with its classical non-full wave form. The Peney landslide (Geneva, Switzerland) is partially vegetated by bushes and trees, and in order to monitor its displacements during the drawdown of the Verbois reservoir located at its base, an alternative solution has been found. We combined LiDAR technique with 14 targets made of polystyrene placed at different locations inside and outside the landslide area. The obtained displacements were compared with classical measurement methods (total station and extensometer), showing good resemblance of results, indicating that the use of targets in highly vegetated areas could be an efficient alternative for mass movements monitoring. [ABSTRACT FROM AUTHOR]

Early warning systems (EWSs) rely on the capacity to forecast a dangerous event with a certain amount of advance by defining warning criteria on which the safety of the population will depend. Monitoring of landslides is facilitated by new technologies, decreasing prices and easier data processing. At the same time, predicting the onset of a rapid failure or the sudden transition from slow to rapid failure and subsequent collapse, and its consequences is challenging for scientists that must deal with uncertainties and have limited tools to do so. Furthermore, EWS and warning criteria are becoming more and more a subject of concern between technical experts, researchers, stakeholders and decision makers responsible for the activation, enforcement and approval of civil protection actions. EWSs imply also a sharing of responsibilities which is often averted by technical staff, managers of technical offices and governing institutions. We organized the First International Workshop on Warning Criteria for Active Slides (IWWCAS) to promote sharing and networking among members from specialized institutions and relevant experts of EWS. In this paper, we summarize the event to stimulate discussion and collaboration between organizations dealing with the complex task of managing hazard and risk related to active slides. [ABSTRACT FROM AUTHOR]

Abstract: Sackung is a widespread post-glacial morphological feature affecting Alpine mountains and creating characteristic geomorphological expression that can be detected from topography. Over long time evolution, internal deformation can lead to the formation of rapidly moving phenomena such as a rock-slide or rock avalanche. In this study, a detailed description of the Sierre rock-avalanche (SW Switzerland) is presented. This convex-shaped postglacial instability is one of the larger rock-avalanche in the Alps, involving more than 1.5billionm3 with a run-out distance of about 14km and extremely low Fahrböschung angle. This study presents comprehensive analyses of the structural and geological characteristics leading to the development of the Sierre rock-avalanche. In particular, by combining field observations, digital elevation model analyses and numerical modelling, the strong influence of both ductile and brittle tectonic structures on the failure mechanism and on the failure surface geometry is highlighted. The detection of pre-failure deformation indicates that the development of the rock avalanche corresponds to the last evolutionary stage of a pre-existing deep seated gravitational slope instability. These analyses accompanied by the dating and the characterization of rock avalanche deposits, allow the proposal of a destabilization model that clarifies the different phases leading to the development of the Sierre rock avalanche. [Copyright &y& Elsevier]

Purpose – The purpose of this paper is to describe empirical research intended to fill the perceived gap in practical guidance methodologies for assessing resilience. Design/methodology/approach – To do so, an interdisciplinary team of researchers studied landslide risk in four different communities of Central and Eastern Nepal using a case study approach. Two case studies on flood-affected communities were developed for comparison sake in more urban areas. Methods combined qualitative participatory approaches to develop indicators of resilience as well as a household survey and focus group discussions for collecting data on the indicators. Findings – What the research results demonstrate is a relatively straightforward and simple means for obtaining data on the state of a community's resilience as a relatively simple "snapshot" of resilience at one period in time, assuming that resilience is an outcome that can be improved over time with the "right" set of interventions. Research limitations/implications – The main limitation of this research is that it focussed mainly on outcome indicators; although some process indicators of resilience were identified (i.e. grazing management practices, skills training, organizational skills and learning through education), these need to be more comprehensive and validated through community consultations. Originality/value – The paper provided data and a straightforward methodology for measuring resilience and has thus contributed to the literature on this topic, while providing practical ideas for future research on resilience building measures and indicators. [ABSTRACT FROM AUTHOR]

Abstract: The Champlain Sea clays of Eastern Canada are incised by numerous rivers. Their slopes have been modified by landslides: on the Chacoura River near Trois-Rivières (Quebec), several large landslide scars, more or less recent, are visible. The role of erosion (channel incision, lateral channel migration and erosion of slopes due to agricultural drainage) as a trigger of these landslides is important. The aim of this study is to understand how erosion and landslides are related to valley development. From a detailed analysis of aerial photographs and DEMs, a map of the phenomena has been drawn by identifying various elements such as landslides, limits of the slope, position of the channel, and the area covered by forest. It is shown that channel change and erosion are strongly linked to landslides by the fact that they change the bank morphology in an unstable way. A slide in itself is a natural way for the slope to achieve stability. But when it occurs in a stream, it creates a disturbance to the stream flow enhancing local erosion which may change the river path and generate more erosion downstream or upstream resulting in more slides. Cross-valley sections and a longitudinal profile show that landslides are a major factor of valley formation. It appears that the upper part of the Chacoura River valley is still unaffected by landslides and has V-shaped sections. The lower part has been subject to intense erosion and many landslide scars can be seen. This shows that the valley morphology is transient, and that future activity is more likely to occur in the upper part of the river. Therefore the identification of areas prone to erosion will help determine the possible location of future large landslides just like the ones that occurred in the lower part. [Copyright &y& Elsevier]

This paper presents a short history of the appraisal of laser scanner technologies in geosciences used for imaging relief by high-resolution digital elevation models (HRDEMs) or 3D models. A general overview of light detection and ranging (LIDAR) techniques applied to landslides is given, followed by a review of different applications of LIDAR for landslide, rockfall and debris-flow. These applications are classified as: (1) Detection and characterization of mass movements; (2) Hazard assessment and susceptibility mapping; (3) Modelling; (4) Monitoring. This review emphasizes how LIDAR-derived HRDEMs can be used to investigate any type of landslides. It is clear that such HRDEMs are not yet a common tool for landslides investigations, but this technique has opened new domains of applications that still have to be developed. [ABSTRACT FROM AUTHOR]

On December 4th 2007, a 3-Mm landslide occurred along the northwestern shore of Chehalis Lake. The initiation zone is located at the intersection of the main valley slope and the northern sidewall of a prominent gully. The slope failure caused a displacement wave that ran up to 38 m on the opposite shore of the lake. The landslide is temporally associated with a rain-on-snow meteorological event which is thought to have triggered it. This paper describes the Chehalis Lake landslide and presents a comparison of discontinuity orientation datasets obtained using three techniques: field measurements, terrestrial photogrammetric 3D models and an airborne LiDAR digital elevation model to describe the orientation and characteristics of the five discontinuity sets present. The discontinuity orientation data are used to perform kinematic, surface wedge limit equilibrium and three-dimensional distinct element analyses. The kinematic and surface wedge analyses suggest that the location of the slope failure (intersection of the valley slope and a gully wall) has facilitated the development of the unstable rock mass which initiated as a planar sliding failure. Results from the three-dimensional distinct element analyses suggest that the presence, orientation and high persistence of a discontinuity set dipping obliquely to the slope were critical to the development of the landslide and led to a failure mechanism dominated by planar sliding. The three-dimensional distinct element modelling also suggests that the presence of a steeply dipping discontinuity set striking perpendicular to the slope and associated with a fault exerted a significant control on the volume and extent of the failed rock mass but not on the overall stability of the slope. [ABSTRACT FROM AUTHOR]

South Peak is a 7-Mm potentially unstable rock mass located adjacent to the 1903 Frank Slide on Turtle Mountain, Alberta. This paper presents three-dimensional numerical rock slope stability models and compares them with a previous conceptual slope instability model based on discontinuity surfaces identified using an airborne LiDAR digital elevation model (DEM). Rock mass conditions at South Peak are described using the Geological Strength Index and point load tests, whilst the mean discontinuity set orientations and characteristics are based on approximately 500 field measurements. A kinematic analysis was first conducted to evaluate probable simple discontinuity-controlled failure modes. The potential for wedge failure was further assessed by considering the orientation of wedge intersections over the airborne LiDAR DEM and through a limit equilibrium combination analysis. Block theory was used to evaluate the finiteness and removability of blocks in the rock mass. Finally, the complex interaction between discontinuity sets and the topography within South Peak was investigated through three-dimensional distinct element models using the code 3DEC. The influence of individual discontinuity sets, scale effects, friction angle and the persistence along the discontinuity surfaces on the slope stability conditions were all investigated using this code. [ABSTRACT FROM AUTHOR]

Copyright of Canadian Geotechnical Journal is the property of Canadian Science Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Canadian Geotechnical Journal is the property of Canadian Science Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this paper, a MATLAB tool for the automatic detection of the best locations to install a wireless sensor network (WSN) is presented. The implemented code works directly on high-resolution 3D point clouds and aims to help in positioning sensors that are part of a network requiring inter-visibility, namely, a clear line of sight (LOS). Indeed, with the development of LiDAR and Structure from Motion technologies, there is an opportunity to directly use 3D point cloud data to perform visibility analyses. By doing so, many disadvantages of traditional modelling and analysis methods can be bypassed. The algorithm points out the optimal deployment of devices following mainly two criteria: inter-visibility (using a modified version of the Hidden Point Removal operator) and inter-distance. Furthermore, an option to prioritize significant areas is provided. The proposed method was first validated on an artificial 3D model, and then on a landslide 3D point cloud acquired from terrestrial laser scanning for the real positioning of an ultrawide-band WSN already installed in 2016. The comparison between collected data and data acquired by the WSN installed following traditional patterns has demonstrated its ability for the optimal deployment of a WSN requiring inter-visibility. [ABSTRACT FROM AUTHOR]

• Self-made mixed label landslide dataset is constructed using open-source optical remote sensing images. • A weight transfer function is designed to enable autonomous learning of mask segmentation information through boundary regression. • Mask score calculation method is proposed to self-correct loess landslide detection mask. • Average precision of loess landslide classification is improved by 20.7%. Loess landslides pose a severe threat of destruction, and detecting them is crucial for minimizing their impact on society. They typically consist of wind-deposited clay and silt, which makes them challenging to detect using conventional methods. Techniques like visual interpretation and field surveys are the most useful, yet these methods can be laborious, expensive, and require a certain level of prior knowledge. Furthermore, remote sensing approaches face the challenge of distinguishing between natural erosion and landslides. Recently, deep learning for landslide detection has the potential to speed up and improve detection accuracy. Nonetheless, deep learning models require large amounts of labeled data and the development of robust algorithms capable of extracting meaningful features from remote sensing data. In this article, a novel approach is introduced to improve the Mask Regional Convolutional Neural Network (Mask-RCNN) algorithm for accurately detecting landslides when the number of available segmentation mask samples is limited. Specifically, A novel loess landslides dataset is established using high-resolution remote sensing images in Gansu Province. In the context of partially supervised learning, a neural network branch containing a weight transfer function is developed to capture mask information from bounding boxes. Additionally, a mask-scoring network block is used to learn the quality of predicted instance masks. Our modified algorithm achieves an average precision improvement of 20.7% compared to the original Mask R-CNN algorithm in small landslide detection. The mask IoU threshold value of 0.5 is used to estimate the average accuracy higher than 0.75. The average precision of the segmentation mask is improved by 16.7% in test set. By proposing a solution that can achieve accurate landslide detection while using limited labeled samples, this study makes a valuable contribution to the application of deep learning in the domains of remote sensing and landslide detection. [ABSTRACT FROM AUTHOR]

Abstract: This study proposes a new concept for upscaling local information on failure surfaces derived from geophysical data, in order to develop the spatial information and quickly estimate the magnitude and intensity of a landslide. A new vision of seismic interpretation on landslides is also demonstrated by taking into account basic geomorphic information with a numeric method based on the Sloping Local Base Level (SLBL). The SLBL is a generalization of the base level defined in geomorphology applied to landslides, and allows the calculation of the potential geometry of the landslide failure surface. This approach was applied to a large scale landslide formed mainly in gypsum and situated in a former glacial valley along the Rhône within the Western European Alps. Previous studies identified the existence of two sliding surfaces that may continue below the level of the valley. In this study, seismic refraction–reflexion surveys were carried out to verify the existence of these failure surfaces. The analysis of the seismic data provides a four-layer model where three velocity layers (<1000ms−1, 1500ms−1 and 3000ms−1) are interpreted as the mobilized mass at different weathering levels and compaction. The highest velocity layer (>4000ms−1) with a maximum depth of ~58m is interpreted as the stable anhydrite bedrock. Two failure surfaces were interpreted from the seismic surveys: an upper failure and a much deeper one (respectively 25 and 50m deep). The upper failure surface depth deduced from geophysics is slightly different from the results obtained using the SLBL, and the deeper failure surface depth calculated with the SLBL method is underestimated in comparison with the geophysical interpretations. Optimal results were therefore obtained by including the seismic data in the SLBL calculations according to the geomorphic limits of the landslide (maximal volume of mobilized mass=7.5×106 m3). [Copyright &y& Elsevier]

[Display omitted] • A new method for slow displacement prediction of reservoir landslide is proposed. • The deformation mechanism of different regions of landslide is different. • The Jiuxianping landslide is affected by the combination of rainfall and reservoir. • The mechanism of step-like slow displacement of reservoir landslide is discussed. • Empirical mode decomposition can effectively improve the prediction model accuracy. Landslides triggered by extreme rainfall due to global climate change are becoming more frequent. The Earth surface processes activity and landform evolution caused by landslide movement seriously affect human survival and the environment. In recent years, the reservoir slope located in the Three Gorges reservoir (TGR) area has been subject to the combined effects of extreme rainfall and reservoir regulation, which have been highly susceptible to inducing landslides. Reservoir landslides usually show step-like slow movement characteristics. The causes of slow-moving landslides are complex, the mechanisms behind them are difficult to grasp, and it may evolve into violent landslide disasters. Therefore, accurately predicting step-like displacement in slow-moving landslides is an effective solution for risk reduction. To achieve the accurate prediction of the step-like displacement of slow-moving landslides and the analysis of the failure mechanism, this study establishes a prediction model applicable to step-like displacement prediction of slow-moving reservoir landslides is developed based on 16 years of continuously monitored cumulative displacement of the Jiuxianping landslide, combined with the Empirical Mode Decomposition (EMD) method and the Improved Particle Swarm Optimization (IPSO) optimized the Long short-term memory (LSTM) neural network. This approach can effectively improve step-like displacement prediction accuracy. In particular, this approach can better represent the association between the step displacement of slow-moving landslides with environmental characteristics (e.g., rainfall, reservoir level). The study reveals that there are differences in the mechanisms and influencing conditions that cause the deformation of different areas of the landslide. Additionally, the reservoir water level fluctuation and seasonal rainfall are identified as the primary reasons contributing to the stepped displacement of the Jiuxianping landslide. The method provides a basis for the step-like displacement prediction, instability mechanism and landform evolution study of slow-moving reservoir landslides and provides valuable guidance for the prevention and control of similar reservoir geomorphic hazards worldwide. [ABSTRACT FROM AUTHOR]

In the morning of 23 August 2017, a maximum of 3.5 million m³ of granitoid rock broke off from the east face of Piz Cengalo, SE Switzerland. The resulting landslide entrained a maximum of 0.8 million m³ of a glacier and continued as a rock avalanche, before evolving into a channelized debris flow which arrived at the village of Bondo at a distance of 6.5 km after a couple of minutes. The event resulted in eight fatalities in the upper part of the valley and led to severe damage in Bondo. The most likely candidate for the origin of the water causing the transformation of the rock avalanche into the initial debris flow surge is the entrained glacier ice. The amounts of water or ice stored in cracks in the failed rock mass and in the deposit of an older rock avalanche beneath are probably much less relevant. Subsequent debris flows can be attributed to stream flow having accumulated behind or leaked into the rock avalanche deposit, and melting of ice in the rock avalanche deposit.We present a set of back-calculations of the process chain of the initial debris flow surge, employing the two-phase and three-phase mass flow models implemented with the r.avaflow computational framework. Thereby, we investigate inhowfar the documentation of the event can be reproduced by the model in a plausible way. Entrainment and melting of the glacier ice as the sole water source can explain the observations. However, due to the relatively low overall ratio between melt water and rock the formation of a debris flow requires some further assumptions. We consider two possible scenarios: (i) a spatio-temporally differentiated water content in the landslide directly after entrainment of the glacier ice can be generated with r.avaflow by assuming entrainment of most of the ice by the frontal part of the initial landslide, leading to a debris flow at the front, with the rear part remaining mostly dry and depositing mid-valley. (ii) Another possible transformation mechanism relies on the assumption that the entrained glacier first remains beneath and then moves behind the rock avalanche as an avalanching flow of water and ice, and partly overtops and entrains the stopping rock avalanche immediately afterwards, generating the first debris flow surge.In summary, our simulations confirm the plausibility of the assumed mechanisms of process transformation. However we note that these results are still preliminary. More research is necessary to better understand the initial conditions and mechanisms of this extreme landslide event. [ABSTRACT FROM AUTHOR]

Abstract An earthflow/mudflow of a few thousand cubic metres occurred at the toe of the Pont-Bourquin Landslide (PBL, Swiss Alps) in late August 2010. This event was preceded by a drop of about 6% in surface wave velocity (dV/V) determined from ambient vibration records. A seismic monitoring system made of three pairs of sensors was re-installed across the transportation and accumulation zones of the landslide in October 2011, allowing daily relative changes in seismic velocity to be measured for a period of 4.5 years. No similar drop in dV/V was observed during this period, consistently with the lack of significant landslide acceleration or earthflow/mudflow events. However, the three dV/V time series showed periodic and reversible variations in a range − 2% to 2%, suggesting a probable influence of seasonal parameters. They were cross-correlated to daily environmental (temperature and rainfall) and surface displacement time series. In the long term (yearly scale), dV/V variations are mainly driven by the temperature with short delays (30 to 50 days) indicating that the shallow layer (first 2 m) controls the dV/V variations. In the short term, the landslide response to precipitations exhibits a small decrease in dV/V with a delay of 2 to 5 days, in contrast with the displacement rate that almost instantaneously responds to the rainfall. The continuous seismic monitoring of PBL using ambient vibrations has proved to be a robust method for getting information at depth, with no data gap even during winters. The seasonal reversible seismic velocity variations turned out to be in a range lower than the drop observed before the August 2010 earthflow/mudflow, highlighting the interest of incorporating the dV/V technique in monitoring systems. [ABSTRACT FROM AUTHOR]

This paper details an investigation of land use/land cover changes (LULCC) for the period 1979–2016 in the Phewa Lake area watershed in Western Nepal based on an analysis of aerial photos, satellite images and ground-based observations. The main objective is to document LULCC with regards to forest, agriculture, built-up areas and roads in order to understand how sources of erosion have evolved during the investigated period. Results demonstrate that LULCC have led to a shift of active erosion areas from grazing zones around streams and forests in the northern part of the watershed to a much greater number of small failures along unplanned earthen rural roads, which have hugely increased in the watershed. The extreme rainfall event that occurred in July 2015 trigged a large number of small and large landslides, as prior to this date only 14 landslides were observed in the watershed compared with 174 after the event. More than 40% of newly triggered landslides intersect a road and 84% of them are located within a distance of 40 meter from a road. Roads are thus influencing slope stability through slope cutting, concentrating surface water, thus changing hydrological patterns, which are destabilizing upper slopes. This study quantifies observable trends of erosion in Phewa watershed, which are important for more sustainable management of rural road construction. Greater enforcement of regulations on road construction and more sustainable land management practices are required to reduce further environmental and economic impacts in Phewa Lake watershed. [ABSTRACT FROM AUTHOR]

Abstract: Landslides are an increasing problem in Nepal’s Middle Hills due to both natural and human phenomena: mainly increasingly intense monsoon rains and a boom in rural road construction. This problem has largely been neglected due to underreporting of losses and the dispersed nature of landslides. Understanding how populations cope with landslides is a first step toward developing more effective landslide risk management programs. The present research focuses on two villages in Central-Eastern Nepal, both affected by active landslides but with different coping strategies. Research methods are interdisciplinary, based on a geological assessment of landslide risk and a socio-economic study of the villages using household questionnaires, focus group discussions and transect walks. Community risk maps are compared with geological landslide risk maps to better understand and communicate community risk perceptions, priorities and coping strategies. A modified typology of coping strategies is presented, based on previous work by that is useful for decision-makers for designing more effective programs for landslide mitigation. Main findings underscore that coping strategies, mainly seeking external assistance and outmigration, are closely linked to access to resources, ethnicity/social status and levels of community organization. Conclusions include the importance of investing in organizational skills, while building on local knowledge about landslide mitigation for reducing landslide risk. There is great potential to increase coping strategies by incorporating skills training on landslide mitigation in existing agricultural outreach and community forest user group training. [Copyright &y& Elsevier]

We introduce a new technique to monitor landslide surface displacements, using radio-frequency identification (RFID) passive tags [1]. Landslide ground deformation monitoring at a local scale requires accuracy, along with dense spatial and time resolution, and must operate in any weather conditions. There are already solutions available to track this motion, such as wire extensometers, differential GPS, optical laser, LIDAR, radar interferometers, or radar nodes. However, none of them offers at the same time a very low cost per point for dense monitoring, a cost-efficient station for studying small sites, and a continuous monitoring under difficult weather such as rainfall, fog, and snow. Previous RFID studies suggest that a dense network of passive RFID tags could be deployed on hectometer-scale landslides at low cost, to monitor 1D radial displacements with an accuracy of 1-2 cm in a variety of meteorological conditions [2, 3] To test the performance of RFID, for five months, we monitored the displacements of 19 tags placed 5 m apart on a medium-sized landslide (Pont-Bourquin, Switzerland) and compared the results to those obtained with total station and extensometer data [1]. The RFID tracking, based on the observation of the phase-difference-of-arrival evolution, showed a better continuity and accuracy in time, compared to the wire extensometer. This was particularly true during rainfall or after a snowfall. In term of performances, the RFID technique reached a general accuracy of 1 cm, and 8 cm after a snowfall, at a distance up to 60 m between the moving tag and the fixed interrogator. In conclusion, RFID displacement tracking appears effective for monitoring surface deformations on a landslide, with unpreceded spatio-temporal sampling, at a significantly lower cost than GPS or ground-based radar and with an overall 1-cm accuracy within a range of 60 m. It appears less sensitive to environmental conditions such as snowfalls compared to classic wire extensometers or optical methods. In the future, RFID tags could lead to new outdoor applications, such as on volcanoes or civil infrasctructure, and exploit their ability for sensing moisture, temperature or vibrations in addition to their displacement. Finally, the fast pace of technological developments in RFID should lead to considerable improvements to the technique in the near future. [1] Le Breton, M., Baillet, L., Larose, E., Rey, E., Benech, P., Jongmans, D., Guyoton, F., Jaboyedoff, M., 2019. Passive radio-frequency identification ranging, a dense and weather-robust technique for landslide displacement monitoring. Engineering Geology. In Press. https://doi.org/10.1016/j.enggeo.2018.12.027[2] Nikitin, P.V., Martinez, R., Ramamurthy, S., Leland, H., Spiess, G., Rao, K.V.S., 2010. Phase based spatial identification of UHF RFID tags, in: IEEE Int. Conf. RFID, Orlando, FL, USA, pp. 102–109. https://doi.org/10.1109/RFID.2010.5467253[3] Le Breton, M., Baillet, L., Larose, E., Rey, E., Benech, P., Jongmans, D., Guyoton, F., 2017. Outdoor UHF RFID: Phase Stabilization for Real-World Applications. IEEE Journal of Radio Frequency Identification 1, 279–290. https://doi.org/10.1109/JRFID.2017.2786745 [ABSTRACT FROM AUTHOR]