GSA Annual Meeting in Seattle, Washington, USA - 2017

Paper No. 354-9
Presentation Time: 9:00 AM-6:30 PM


MEGHANI, Nooreen A., Earth and Environmental Systems Institute, The Pennsylvania State University, 318B EES Building, University Park, PA 16802, MILLER, Douglas A., Earth and Environmental Systems Institute, The Pennsylvania State University, 317G EES Building, University Park, PA 16802 and HOLDERMAN, Brennan, Ecosystem Science and Management, The Pennsylvania State University, University Park, PA 16802,

Small, unmanned aerial systems (sUAS), commonly called “drones”, are becoming increasingly valuable in the geosciences. Much of the data that were previously collected through expensive and time-consuming manned aerial missions, repeated over decades, can now be gathered with today’s lightweight, easy to use, and relatively inexpensive sUAS. Off-the-shelf drones come equipped with excellent cameras (for example, the DJI Phantom 4 Advanced and Pro models have cameras with 1” sensors and cost $1300 and $1600, respectively), and automated photogrammetry software have allowed sUAS photogrammetry to become commonplace. Photogrammetry software can produce point clouds comparable to terrestrial lidar (Hugenholtz et al. 2013), and digital surface models (DSM) with internal (within-image) accuracies on the order of centimeters (James and Robson, 2012). Images that are orthorectified based solely on the camera’s internal GPS have similar over-all accuracy as consumer-grade hand-held GPS units and can suffer from both rotational and translational inaccuracies. Including surveyed ground control points (GCP) and identifying them in photogrammetric software improves accuracy dramatically (e.g. Aguera-Vega et al., 2017); however, placing GCP for use in sUAS missions is not always practical: densely forested areas are one such example. Even in leaf-off conditions, accuracy achieved by survey-grade GNSS systems is worse than 2m (US Forest Service GPS systems review, as of June 8, 2017) and cost tens of thousands of dollars. For less than $600, Emlid offers real time kinematic (RTK) GNSS sensors (currently L1 only, L1/L2 in the next model) that are reported to get 2cm accuracy or better (Reach RTK specs). Using one Emlid receiver as a base station, set up either on a forest service road or in a clearing, and another Emlid receiver as a rover on a T18 Octocopter, we aim to overcome the need for GCPs by instead more accurately locating our sUAS. Initial stationary tests with the Emlid system have resulted in better than 30cm horizontal accuracy with the base station and better than 5 cm horizontal accuracy with the rover in relation to the base station. These preliminary results suggest that we will be able to achieve sub-meter accuracy in our aerial surveys with the Emlid Reach RTK system.