LiDAR as input for Discrete Fracture Networks: A comparison of automated manual joint mapping using scanned surface models.

I. Vazaios, N. Vlachopoulos, M.J. Lato, M.S. Diederichs

In the proceedings of: GeoRegina 2014: 67th Canadian Geotechnical Conference

Session: Engineering Geology and Rock Mechanics

ABSTRACT: LiDAR technology has been proven to be a real asset in many geological and geotechnical applications but it can also be a powerful tool for modelling. Discontinuity data, such as joint orientation, density and length, can be extracted from LiDAR scanning data using manual and automated techniques. Such data can be further processed and used in DFN modeling and generation. However, the sampling method applied in the LiDAR data and the reconstruction method of the DFN, including deterministic and statistical approaches, may result to realistic or unrealistic DFN models. Different methodologies and techniques are discussed and compared based on two specific case studies, including data sets obtained from the Shawinigan, Quebec, Canada and Brockville, Ontario, Canada railway tunnels. RÉSUMÉ La technologie Lidar a démontré qu'elle peut être un véritable atout dans de nombreuses applications géologiques et géotechniques, mais elle peut également être un outil puissant pour la modélisation. Des données discontinues, comme l'orientation des joints de stratification, la densité et la longueur, peuvent être extraites à partir de données obtenues grâce à des balayages manuels ou automatiques. Ces données peuvent être traitées et utilisées pour la modélisation d'un réseau de fractures discrètes. Cependant, la méthode d'échantillonnage de données utilisée par la technologie Lidar et la méthode de reconstruction du réseau de fractures discrètes, y compris les approches déterministiques et statistiques, peuvent entraîner des modèles réalistes ou irréalistes. Différentes méthodes et techniques sont discutées et comparées selon deux études de cas spécifiques, qui incluent des données obtenues à partir des tunnels ferroviaires de Shawinigan, Québec, Canada et de Brockville, Ontario, Canada. 1 INTRODUCTION LiDAR (Light Detection and Ranging) technology has been proven to be a real asset to many industry sectors as a response to the growing need for rapid, cost effective and accurate generation of relative geospatial information. More specifically in underground projects, like tunnels, LiDAR enables the user to scan the excavated area in order to produce accurate 3D surface models. The data which is accumulated provide useful information regarding the structural features of the rockmass, as the discontinuities intersecting the tunnels perimeter can be mapped within the LiDAR data instead of applying traditional measurement methods which are especially time consuming and result in a smaller measurement sample. Such data sets can either be used as input into a statistical rockmass model or to create a discrete replica of the rock structure as observed in-situ. During physical mapping selected features can be missed due to the inherent limitations of the process. Even manual mapping of a virtual LiDAR model can result in subjectivity in data selection. In order to facilitate and accelerate such assessments, automated systems have been developed (Lato and Vöge, 2012). However, they can result in noisy data sets with numerous false indications of structures or may result in missed data, corresponding to real discontinuities, which fail to satisfy the search criteria. Either method, when associated with the generation of Discrete Fracture Networks (DFNs), can produce unnatural or unrealistic synthetic joint networks, if due care and adequate verification measures are not taken. Another significant factor in generating realistic and reliable DFNs is the reconstruction method applied, i.e. including a deterministic and/or a statistical method (Fekete and Diederichs, 2012). The deterministic approach produces a rockmass model constructed from discretely measured discontinuities in LiDAR data, defining each unique planar surface by a specific orientation and spacing measured in situ. On the contrary, the statistical approach is based on the interpretation of the joint sets' orientation and spacing measured in-situ, highly depending on the mathematical assumptions and thus resulting in the generation of multiple joint networks which are not necessarily realistic. Either approach requires validation in the context of the site-specific geological history and geological features, in order to provide optimal results. In this paper, the aforementioned methodologies and techniques will be discussed and compared based on two specific case studies, including data sets obtained from the Shawinigan, Quebec, Canada and Brockville, Ontario, Canada railway tunnels. 2 LIDAR DATA LiDAR uses transmitted and reflected laser signals resulting in the recording of millions of points of high accuracy in space, referred as the point cloud. From the

RÉSUMÉ: as input for Discrete Fracture Networks:

Access this article:
Canadian Geotechnical Society members can access to this article, along with all other Canadian Geotechnical Conference proceedings, in the Member Area. Conference proceedings are also available in many libraries.

Cite this article:
I. Vazaios; N. Vlachopoulos; M.J. Lato; M.S. Diederichs (2014) LiDAR as input for Discrete Fracture Networks: A comparison of automated manual joint mapping using scanned surface models. in GEO2014. Ottawa, Ontario: Canadian Geotechnical Society.

@article{GeoRegina14Paper331,author = I. Vazaios; N. Vlachopoulos; M.J. Lato; M.S. Diederichs,title = LiDAR as input for Discrete Fracture Networks: A comparison of automated manual joint mapping using scanned surface models.,year = 2014}