Automated Methods for Fully Exploring and Interpreting LIDAR Data Points

The LIDAR scanning combined with digital photograph mapping techniques (Bellian et al. 2005) has become a privileged tool to obtain a 3D georeferenced reconstruction of an outcrop, often termed as DOM (Digital Outcrop Model). For several years, many geoscientist applications use DOMs as support for manual interpretations of strata or fractures and facies mapping. However, the LIDAR tool produces huge data sets that become easily difficult to manipulate interactively and then to interpret. A new challenge in geomodelling is then to extract, in an automated way, geological features from a DOM. Different kinds of strategies have been proposed in the litterature based on both LIDAR points or DOMs. For example, some authors have proposed to use maximum curvature values (Ahlgren et al., 2003) in order to obtain statistics about fracture networks (orientations, density). Automated detection methods have been presented in Kudelski et al. (2009), Viseur (2008) and Viseur et al. (2009). They are applied on DOMs and they aim at extracting as polygonal lines the strata or fracture paths observed along the outcrop. Other authors (Garcia-Selles et al., 2008; Franceschi et al., 2009) use properties computed (geometrical attributes) or available (intensity) from LIDAR data points to highlight or detect geological features. Finally, authors have proposed approaches based on the ”ant tracking” algorithms applied on the colours of the mapped pictures (Monsen et al., 2007). In this paper, a series of algorithms are presented. They are integrated into a workflow to fully explore and interpret numerical outcrops from data points to horizon or fracture surface constructions. Indeed, working on DOMs requires to build surfaces from very dense multivalued XYZ data points which is time consuming and generally leads to mesh decimation in order to obtain triangulated surfaces light to manipulate. These operations may damage the information contained in the topography geometry. Therefore, working directly onto the XYZ data points may be a good alternative and allows the display of subtle relief signals. Moreover, the LIDAR engin is experiencing new developments and LIDAR data points with RGB flags are increasingly provided. The proposed approach aims at first extracting as polygonal lines the limits of geological objects from the LIDAR data points. Then, surfaces may be built to model the detected fractures or strata interfaces.