ACCURACY ASSESSMENT OF SUAS- AND LOW-COST- GNSS ALONG A FORESTED ROAD IN CENTRAL PENNSYLVANIA

Survey-grade GNSS equipment often performs poorly in densely forested regions due to canopy occlusion. The development of low-cost ‘mapping grade’ (better than decimeter accuracy in ideal conditions) GNSS systems allows for similar accuracy in these locations at a fraction of the cost of survey-grade equipment. Drones can collect GPS data above the canopy, but cannot perform long-term occupations of a single location. However, sUAS and low-cost GNSS systems used in tandem may be able to provide an inexpensive solution for higher accuracy in forested regions.

We installed twelve ground control points (GCPs) with 8” aerial markers at clearings along 2 km of forested road in central Pennsylvania. GCPs were occupied for 10-15 minutes each with an Emlid REACH GNSS receiver mounted to a Leica survey tripod with optical plummet. Data were post-processed against CORS station PSU1, 15 km north, with open-source RTKlib software. We collected geo-tagged aerial imagery at 300 feet above ground level with a DJI Phantom 4 advanced and built-in 20 MP camera. We used Agisoft Photoscan to produce an orthomosaic from 993 photos and extracted the raw location data of each GCP for comparison to the Emlid data.

To assess the accuracy of the Emlid and Phantom data the GCPs were also occupied by survey-grade GNSS equipment. We assume the survey-grade equipment to have the highest accuracy and make all of our comparisons to those data.

The horizontal positions of the GCPs extracted from the raw orthomosaic are within 2 m of the post-processed Emlid GCPs, and both datasets are typically within 2 m of the survey-grade data. The highest quality Emlid points are within 10 cm of the survey-grade points. The raw orthomosaic shows little topographic change while the Emlid shows nearly 50 m of change, in good agreement with the survey-grade data. Nine of the Emlid points are within 1 m elevation of the survey data, three are within 10 cm. Using the three highest quality Emlid points to georeference the orthomosaic dramatically improves the topographic representation over the entirety of the road segment.

These preliminary results suggest that a combination of low-cost GNSS and aerial imagery can yield significantly higher quality spatial information in forested regions than can be achieved with terrestrial GPS equipment alone.