EVALUATION OF PPK-CORRECTED GEOTAGS FOR THE REDUCTION OR ELIMINATION OF GROUND CONTROL POINTS IN FOREST AND PASTURE SETTINGS

In Structure-from-Motion photogrammetry (SfM) projects utilizing drone imagery, a time consuming and costly step of the data collection process is the deployment of ground control points (GCPs). GCPs provide known georeference points within the imagery acquired by the drone. This approach (i.e., georeferencing based on the content of the imagery) is typically used because it does not require the position of the camera to be precisely known at the time each photograph is taken. In theory, georeferencing could be achieved by knowing the precise location and orientation of the camera at the exact moment each photograph was captured. A new generation of GNSS positioning devices allow accurate and precise camera coordinates to be captured and recorded on board the drone simultaneously to the capture of each photograph. These systems will be tested with visible spectrum photos to determine if the accuracy and precision are adequate to reduce or eliminate the requirement for GCPs, allowing for more rapid drone data acquisition and imagery post-processing.

Fellows Lake, a reservoir north of Springfield, Missouri, was chosen as our 0.52 square kilometer study area as the lake provides a range of elevations representative of typical topographic relief in the midcontinent as well as including typical vegetative covers. Data was collected with a Reach M+ GNSS module installed in a DJI Phantom 4 Professional via a third party a Teokit integration kit utilizing an AShot photo board. This upgrade to the drone allows a timemark to be capture in the exact middle of the photo exposure along with UAV telemetry information. We used this data to determine precise camera coordinates based on PPK post-processing of the GNSS raw data log. The corrected GNSS coordinates for each photo were then used during SfM processing to georeference the resulting DSM and Orthorectified mosaic. A partner study found RMSEs for 0 and 5 GCPs to be 158.37 and 2.87, respectively, and MAEs for 0 and 5 GCPs to be 158.36 and 2.51 respectively. Preliminary data in this study found RMSEs for 0 and 5 GCPs to be 7.71 and 7.71, respectively, and MAEs for 0 and 5 GCPs to be 5.378 and 5.383 respectively. As such it appears the number of required GCPs is between 0 and 5.