Comparing manual and remote sensing field discontinuity collection used in kinematic stability assessment of failed rock slopes

Data collection for kinematic and structural analyses in hills or rock slopes still maintains, nowadays, an important manual component. The set of lithological, structural and characteristic observations of the rock mass, directed to engineering design, is referred to as geomechanics observation point or geomechanical station. The properties and orientation of discontinuities and rock matrix of a geomechanics observation point are obtained in situ, by normalized templates. Subsequent laboratory/office work determines the resistant properties of the discontinuities, evaluates the geomechanical quality of the rock mass (e.g. using the RMR, Q, or GSI indices) and carries out a census of discontinuities, grouping into main sets. Field data are combined with laboratory results, in situ and geophysical tests, with the objective of elaborating a "geomechanical model" of the rock mass. Field characterization plays an important role in preliminary studies and, in many cases, is the only information available in the first stages of investigation and project development. Traditionally, rock mass field data collection has been carried out by physically accessing the slope. However, remote acquisition techniques have provided a new perspective. The main two remote data collection techniques are Interferometric Synthetic Aperture Radar –InSAR-and Light Detection and Ranging -LiDAR, InSAR enables high precision measurements of terrain surface movements, and is usually applied to monitor and detect landslides. LiDAR provides a 3D point cloud of the terrain surface. The comparison of point clouds acquired in different time frames also enables the monitoring and detection of landslides. Additionally, the use of 3D point clouds offers the possibility of working with geometrical surface information. Complementarily, digital photography has experienced significant advances with the development of photogrammetry techniques. The Structure from Motion (SfM) technique must be highlighted, as it enables surface reconstruction from 3D point clouds of digital photographs. SfM is considered a high-resolution, low-cost automated photogrammetry method, and it is based on the same principles of stereoscopic photogrammetry (the 3D structure can be built from image superposition). SfM originated from the artificial vision area and from the development of automated algorithms for digital image correlation (DIC). SfM is different from conventional photogrammetry, as the scene geometry, camera positions and orientation are solved automatically without the need of establishing, a priori, a control point network with known 3D coordinates. Collinearity equations are solved from the high number of conjugated points (common image points) identified during the automatic correlation phase of a set of superimposed images acquired in an unstructured way. The equipment utilized in the application of this technique entails lower economic costs than with LiDAR instrumentation, and results are reasonably acceptable. The limitation of SfM is mainly the dependence on the quality of the lens and the processing time, the procedure for capturing images and resource consumption by the machine. Generally, the higher the number, quality and resolution of the images, the higher the quality of the generated model. This, however, entails in a higher consumption of resources and higher computational time 3D point cloud treatment enables semi-automatic identification and extraction of information from plane discontinuity sets. Such information is very important for the development of kinematic analyses of block-controlled instabilities. The SfM technique9 has been attracting attention and becoming popular, as indicated by the increasing number of publications. These techniques permit the characterization of important geomechanical parameters such as orientation, roughness and persistence and spacing of discontinuities. Of all geomechanical parameters that can be extracted from 3D point clouds, analysis of plane discontinuities has probably been the most approached subject to this date. All proposed methods analyze the surface of the slope from 3D point clouds or from reconstruction via triangular elements (TIN, Triangular Irregular Network). In each surface point or element, the normal vector is estimated and a pole is determined; as a result, there is a population of poles available to analyze predominant surface orientations. The overarching objective of this work is to compare the results obtained in the collection of discontinuity orientations using a low-cost remote technique (i.e. SfM) and manual collection (compass). To this end, orientations are gathered with a compass for all accessible discontinuities of a slope, and from all discontinuities that are well represented in the 3D point cloud obtained with SfM. Plane extraction and 3D model orientations are carried out in two different ways: (1) by adjusting the planes to the point sets that belong to the same discontinuity, with Cloud Compare software27; and (2) semi-automatically, with Discontinuity Set Extractor software, DSE. The focus herein is only on the collection of discontinuity orientations, not on rock mass properties.Instituto Geológico y Minero de España, EspañaDepartamento de Ingeniería Civil, Universidad de Alicante, Españ