Acquisition, Processing, and Visualization of 1,000 Square Kilometers of Airborne Lidar, Guadalupe Mountains of Texas and New Mexico

Abstract

Scanning airborne lidar instruments are capable of rapidly collecting multi-gigabytes of digital terrain data. Sample density of the terrain is dependent upon acquisition parameters such as instrument pointing direction, laser firing repetition rate, aircraft elevation, and air speed. The Guadalupe and Delaware Mountains Airborne Lidar Project (GDALP) collected laser point data in both downward-looking and side-looking acquisition orientations. The result of this mixed-acquisition orientation was particularly effective in steep canyons targeted for geologicaly significant exposures. The mixed acquisition oritentation reduced data shadows (Bellian et al. 2005) also informally referred to as “holidays,” thus providing relatively continuous point coverage for vertical cliff faces and horizontal surfaces. The limitation of mixing these two survey styles over such a large study area is large file size. Resultant raw data volume from the GDALP survey was over 4 terabytes and processed point cloud data over the ~1000 km2 area of interest (figure 1) is nearly 300 gigabytes. Even with very high-end personal computer “workstations” that have 40+ gigabytes of RAM and state of the art graphics cards, it is not possible to visualize these points at one time. It was therefore necessary to explore non-standard methods of lidar data visualization to facilitate geological interpretation on the entire dataset with a minimum number of subprojects to collate. Methods previously used for creating normals to points (Bellian 2009) were employed for the GDALP survey, albeit on a significantly larger scale than has previously been documented. This methodology combined with parallel processing on the 256 node cluster at the Bureau of Economic Geology, the University of Texas at Austin helped create four subregions from the GDALP survey. These subregions were then able to be displayed with sun-shading applied to facilitate geological interpretation (figure 1). Viewed independently, the resultant four pointclouds were able to be displayed with minimum rendering rendering latency.