Experience has shown that the precision or relative accuracy of a UAV survey after data processing should be around 1~2 pixels in plan and 2~3 pixels in height. The absolute accuracy of a UAV survey will depend only to a small extent on the quality of the georeferencing supplied by the on-board GNSS, and to a much greater extent on the accuracy with which the Ground Control Points (GCPs) are surveyed. UAVs are not normally equipped with high-precision geodetic GNSS, so GCPs must be used to conform the final orthomosaics and DSMs; without them the accuracy of the resulting survey data can suffer a global shift of anything up to 20 metres, depending on atmospheric conditions and satellite geometry.
Before commencing a survey flight, we lay out a sparse grid of GCPs - usually no more than six or eight are needed, even for a large site - and survey their positions using a GNSS RTK survey instrument. By using a sufficient number of GCPs the accuracy of the survey can be derived from the average Ground Sampling Distance (the distance on the ground between pixel centres measured on an image). Using this principle, if the GSD of a survey is 5cm the accuracy could be assumed to be approximately ±5~10cm in plan and ±10~15cm in height. We routinely acquire photogrammetry using a GSD of 1cm, at which point the survey accuracy becomes effectively limited by the GCP survey accuracy.
- Data accuracy using UAV photogrammetric surveys compares very well with traditional land surveys
- Using a UAV can provide a richer dataset and representation of the site which can be as accurate as RTK GNSS
- A UAV photogrammetric survey can offer the most spatially accurate and cost-effective aerial survey data currently available
- The photogrammetry software automatically generates a quality report to demonstrate the accuracy of the DSM