ABSTRACT:
The ability of laser scanners to rapidly provide structural data in high inaccessible rock slopes has been frequently and correctly documented as an important advantage of the technique. Researchers have also shown the potential for characterizing varied rock mass characteristics including discontinuity orientation, spacing, persistence and roughness. In this paper, the authors, using laser scans of several rock slopes, illustrate the limitations and bias in structural data collection that must be allowed for. Orientation bias can occur both horizontally (laser plan angle bias) and vertically (laser elevation bias). It is suggested that the elevation angle of the laser scanner may produce point clouds that, when processed to derive structural data, may at best underestimate or even completely miss critical joint sets. Censoring can result in underestimating discontinuity size. Limitations related to rock mass reflectivity and textures are also highlighted.
1 INTRODUCTION
Terrestrial laser scanning is being increasingly used in rock slope engineering. Recent developments have seen the introduction of point cloud processing software to allow the determination of discontinuity characteristics (Kemeny & Donovan 2005, Monte 2004, Slob et al. 2005). Researchers have shown the potential for characterization of discontinuity orientation, spacing, persistence and roughness (Feng & Roeshoff 2004). The ability of laser scanners to rapidly provide structural data in high inaccessible rock slopes has been frequently and correctly documented as an important advantage of the technique. In contrast, relatively few published studies exist outlining the limitations of laser scanning in rock slope investigations. Bias related to conventional sampling techniques for discontinuity characterization has been widely documented (Terzaghi 1965, Priest 1993, Zhang & Einstein 1998). Sources of bias include: - Orientation: discontinuities sub-parallel to the sampling line or window are under-represented - Censoring: discontinuities larger than the sampling window are not completely characterized - Truncation: discontinuities smaller than a certain size are neglected
