CREATING ACCURATE DIGITAL SURFACE MODELS USING UAV MOUNTED COMMERCIAL GRADE CAMERAS: A PRACTICAL METHODOLOGY TO GENERATE MAP QUALITY GEOREFERENCED IMAGERY AND TOPOGRAPHY

The use of UAVs in earth and environmental sciences has expanded considerably in the last 5 years. Their use is projected to increase as a low-cost alternative to small piloted aircraft-based operations where cost or geography restricts options available to collect reliable high resolution aerial imagery and topography. With a well-planned ground survey component, off-the-shelf commercial cameras mounted on UAVs offer the capability, with post-processing, to provide hi-accuracy, hi-resolution, digital surface models from the survey/camera geometry and orthomosaics.

We have demonstrated this capability using a sample of UAV based imagery of the Djarrit section of Majuro, Republic of the Marshall Islands collected in September, 2016. The collected imagery (537 vertical photographs) was processed and refined to produce a bare-earth digital elevation model with a horizontal RMSE of 0.69 m and a vertical accuracy of 0.3 m (L.E. = 0.28 m) and final orthomosaics with a pixel resolution of resolution of 0.25 m. This level of accuracy was achieved using an integrated ground-aerial planning methodology that included placed ground targets, survey-grade and kinematic GPS surveys and an iterative post-processing workflow that moved away from relying on inaccurate UAV-based GPS. We will present these results and outline a field-work plan and methodology that we hope will aid other researchers.