VGC 2021 conference

From 29th September to 1st October 2021 took place the 4th edition Virtual Geoscience Conference. Some of us seized the opportunity to present there ongoing work and project.

A- Lidia Loiotine : Multi-disciplinary approach for stability analyses in discontinuous rock masses by means
of conventional geostructural-geomechanical surveys and remote sensing techniques

Loiotine L., Andriani G.F., Derron M.H., Jaboyedoff M., Lollino P., Parise M.

In the last two decades, the application of remote sensing techniques in Earth Sciences has become mainstream. As regards the characterization of rock masses, several methods for the semi-automatic and /or automatic extraction of discontinuity sets and their properties, from raw point clouds or processed surfaces, were introduced in the literature. However, when dealing with low-relief outcrops or man-made excavations, extracting discontinuities from point clouds can be challenging because of the lack of exposed surfaces, thus leading to not realistic rock mass characterization. In these circumstances, 2D quantitative analyses on discontinuities mapped in orthophotos seem to have higher reliability. In addition, numerical stability analyses on rock masses often require the simplification of the model to avoid time-consuming computations and non-convergence issues, which results in partly disregarding the remote sensing data. 

Aiming at a full exploitation of the potential of remote sensing techniques in the framework of rock stability analyses, we developed a multi-disciplinary approach which combines conventional geostructural-geomechanical surveys and remote sensing technologies. We carried out Unmanned Aerial Vehicle (UAV) surveys on a 20 m high cliff and on an adjacent low-relief area.  Successively, we applied Structure from Motion technique to obtain a mesh of the cliff and a high-resolution orthophoto of the flat area. We compiled a specific Matlab routine to semi-automatically identify and characterize the discontinuity sets from the joint traces on the orthophoto. Conventional geostructural-geomechanical surveys were performed to validate the results of 2D quantitative analysis of the discontinuities and to assess the non-geometrical parameters (i.e. roughness, wall strength, aperture). Furthermore, we collected representative rock samples for the physical and mechanical characterization. The different lithofacies were mapped on the mesh of the cliff with the help of UAV-acquired photographs.

We then used the above data to perform a 3D Finite Element Analysis, using a software allowing to implement ubiquitous joints. The geometric model was created by importing the Digital Terrain Model (2 m resolution) of a wider zone and merging the high-resolution mesh of the case study. The lithofacies were created by dividing the 3D model with planes generated along the previously mapped interfaces. The physical and mechanical properties of the rock materials, as well as the strength criteria, were defined according to the outcomes of the geotechnical characterization. Moreover, 3 discontinuity sets were added using the 2D quantitative analysis results. To derive the initial tensional state, we created an additional volume which was eroded during the middle-Pleistocene sea-level regression. After assigning the restraints to the model, an appropriate mesh was created. The transition from shallow-water carbonate platform to the current morphology was simulated by means of successive excavations during the computation.

Despite the presence of the weak Plio-Pleistocene transgressive deposits at the top of the cliff, local instabilities were detected along overhanging blocks and karst caves in the jointed limestones and dolostones. The failure mechanisms are compatible with the results of on-site geomechanical investigations.

B- Charlotte Wolff : Monitoring strategy for movements assessment in a challenging remote area: Case study of Cima del Simano (Ticino, Switzerland)

Wolff C., Derron M.H., Choanji T. , Pedrazzini A., Jaboyedoff M.

Steep slopes or fractured rocks near vulnerable elements sometimes require monitoring to detect centimetric to millimetric movements aimed at preventing dangerous rockfall events. Sometimes those hazards are challenging to monitor due to a difficult access to the site (high altitude, snow cover or incised valley). To overcome those difficulties, several remote techniques can be applied. Here we present the case study of the mountain Cima del Simano in the Ticino canton in Switzerland. This is a mountain located in the incised valley of Blenio, whose top reaches 2500m height. The upper part is covered by snow half of the year and is difficult to access without a helicopter. The East part of the mountain presents important open fractures that have been periodically monitored since 2006. To evaluate the risks related to this instability, different remote techniques have been coupled to locate the areas in movement, their average speed and acceleration as well as their susceptibility to failure. Among the techniques used, some can detect centimetric movements (Lidar, DGPS, images co-registration and correlation) while the others detect millimetric movements (Ground-based Insar and satellite Insar, extensometers).

The first step to apply this monitoring strategy is to decide where to install the different devices and the frequency of measurement acquisitions for each technique. For instance, the Ground-Based radar should be closer than 4km from the target area to decrease atmospheric biases  and its direction should not be too perpendicular to the target slope to avoid foreshortening and layover effects and increase the resolution. Thus, we developed a Matlab routine aimed at guiding the user in the selection of the best radar location among several location possibilities.  

After conducting a preliminary review of the different remote sensing techniques, their specifications and limits, we selected the most appropriate deployment of those techniques  to monitor Cima del Simano in order to overcome acquisition difficulties.

C- Cristiano Gygax: A DIY Arduino based low-cost and short-range terrestrial laser scanner

Gygax C., Derron M.H., Jaboyedoff M.

A first prototype of a low-cost terrestrial laser scanner has been developed based on Arduino technology.  Electronics and mechanical components were partly ordered and partly 3D printed, for a total cost of around USD 350. Conceptually, the operation of the device is simple: two stepper motors drive a laser sensor on two axes (horizontal and vertical), and a distance measurement for each of the motors positions is taken. These components are controlled by an Arduino Mega 2560, a powerful microcontroller known for its simplicity and versatility, which also receives the measurements and stores them on a SD card. A smartphone application was also developed to send scanning parameters to the LiDAR via Bluetooth. This first prototype detects on average 200 points/second at a maximum distance of about 200 m with an average error of 2 cm and a maximum resolution of less than 0.012° (1 point every 2.9 mm at a distance of 15 m). The laser spot diameter is about 30 cm at 50 m range.

Tests have been achieved indoors to compare with high-end commercial LiDAR and SLAM scanners. Measurements errors, noise, effects of surface reflectivity, range and incidence angle are assessed on objects whose geometry are known. Then, the device was tested on a real fast-moving rock slope to detect changes.

This device has two main possible applications: 1) for continuous monitoring in areas where the probability of destruction is too high to put a commercial device thousands of time more expensive. In such situation, it can be programmed to make a scan every hour; 2) for educational purpose, a DIY procedure is proposed to build such a device. That can be used by students in geosciences to understand the principles of laser scanning. Present developments include the coupling of a small solar panel, so that the device can be tested in the field on several consecutive days, and a communication module to send the data.

Method to Estimate the Initial Landslide failure Surface  and Volumes using Grid Points and Spline Curves

Michel Jaboyedoff, Gautam Prajapati

Here we present a new method to estimate the Landslide failure surface and volume using the grid points and spline curves based on a tangent of the failure surface. The Model requires fundamental data inputs, which are readily available online. The challenge in landslide hazard assessment is to estimate the volume involved in slope movements before carrying out a detailed field survey. The volumes are usually only defined once the instability is detected. Digital Elevation Models (DEM) ‘s increasing availability enables us to estimate the volumes involved in slope movements, thus limiting expensive field investigations. This Model requires a DEM of the study area and a KML file of contour limits of the landslide as input. The calculation procedure is simple-using DEM; we have made grid points for each cell. The cross-section for each point has been drawn perpendicular to the slope line joining the Highest elevation point and lowest one within the contour limits. A cubic spline curve requires four parameters, which can be taken from each cross-section endpoint (Coordinates of two endpoints and the first derivative at these points).  The Z value has been calculated at each grid point using these parameters, and finally, a 3D failure surface has been generated. And volume can be calculated using the elevation difference of the failure surface and original surface multiplied by the cell size of each grid point.  Approximately 1100-line MATLAB code has been written to execute this process automatically. We have proposed three methods for variation of the angles between the failure surface and the cross-section lines.

In Method-1, we have assumed the angles are constant for each cross-section. In the 2nd Method, the weights for the Angle variation have been given based on the distance of endpoints from the elevation line joining the highest and lowest points. Here the Dip and strike angles vary non-linearly and inversely proportional to the distance. In the 3rd Method, it has been assumed that the failure surfaces’ left and right profiles are part of a big arc, and the Dip and Strike angles vary linearly.

We have tested the Model on the Kotropi landslide of August 2017, Himachal Pradesh, India, and The La Frasse landslide in the Canton of Vaud (Switzerland). The Kotropi landslide is found at N31° 54′ 39.4″ and E76° 53′ 27.4 on the Indian toposheet Number 53A/13. A large scrap of the slope has been washed away along with the vast volume of debris and other properties like two-state government buses and a few other vehicles. Government reports suggest that there have been at least 46 deaths caused by this disastrous event which buried nearly half a kilometre of the highway, disrupting state transportation very severely. We have also shown the 3-D Model of the calculated failure surface and the original surface. Here method-1 has been used with-(Right profile constant Dip=42⁰  and Left profile constant Dip=59º. The Volume has been calculated, and it is approximately 5944193.76 meter³. The implication of this model is straightforward. We can easily get the required data with very little fieldwork, and the processing will hardly take a few minutes to produce the result. There are so many parameters; we can manipulate them accordingly to get better results. We can enrich the real scenario by making more constraints on the variation of the cross-section angle based on field measurement.  This model is also helpful in predicting the area that can be affected by the mass of the failure surface, and we can make a predefined risk zone to avoid any casualties in future.