Leveraging LIDAR-derived Point Clouds for Topographic Characterization

Velásquez, C. F., Glenn, N. F., Ames, D. P. Acquisition of accurate x, y and z coordinates of terrain information throughout large geographic areas is possible with Light Detection and Ranging (LiDAR) technology. The mass of points (or point cloud) remotely acquired with airborne LiDAR sensors, for instance, are used for the creation of quality digital models of terrain. Extraction of hydrological features such as channel networks from digital terrain models is a common task. This task is generally conducted by means of implementing software tools that analyze raster structures for representation of terrain surfaces. Exploring the feasibility of directly and effectively extracting land cover features from a point cloud deserves due attention given both the proliferation of LiDAR data availability and the potential for the information to overcome some disadvantages that are intrinsic to the raster structure. Pursuing an alternative mode to leveraging LiDAR data for hydrologic applications is, therefore, the subject motivating the current research. A methodology for extraction of concave-shaped terrain surfaces is proposed. These features, relevant to hydrological applications, are extracted by means of implementing a convolution operation that is conducted on the x, y, and z coordinates of the point cloud and with an adapted construct for a zero-sum kernel function. The mass of points that are topographically lower of what is referred to as a zero-crossing level appear to reasonably well delineate concave-shaped landforms. The proposed method is, therefore, of merit to further investigation.