The use of onboard UAV GNSS navigation data for area and volume calculation

Photogrammetric models constructed from unmanned aerial vehicle (UAV) data are nowadays used for determining areas or volumes in many scientific and practical applications. The presented study discusses the possibility of direct georeferencing of photogrammetric models based on onboard navigational GNSS. It is widely expected that all UAV-derived data need to be georeferenced with a sub -decimeter accuracy. The aim of the work was to test whether it would not be enough to georeference photogrammetric models only using navigation GNSS. Imagery from several locations was used for the construction of photogrammetric models using GCPs or navigational GNSS only. All models were transformed into the same coordinate system using the same transformation key to facilitate comparisons. For the first comparisons, a 7 -element transformation was performed to facilitate analysis of the systematic shift, rotation, and scale change between the models. It confirmed that the scale change of the photogrammetric model constructed using navigation GNSS data only from that constructed using GCPs was very small, with a maximum change of 2% (scale change of 0.98). The 7-element transformation also revealed that the models are mutually shifted and rotated. The tilts were below 2?, and the horizontal shifts are consistent with the (in)accuracy of the navigational GNSS (the highest deviation was 3.7m). The data were also transformed using a 4 -element transformation (shifted and rotated along the Z-axis) to enable an analysis of the change of the shape. Subsequently, areas and volumes of both point clouds were calculated; the differences in volumes were below 10%, differences in areas even below 4%. For analysis of local deformations on an extensive site, an airfield area was used. The analysis of distances measured on an orthophotomosaic of a relatively large area (9.5 km2) revealed a mean absolute deviation between the GNSS and GCP data, which is a surprisingly good result. This method is suitable for identifying potential local deformations that can occur in photogrammetric models in the spaces between GCPs. The reported experiments demonstrate that the calculations of volumes and areas using only onboard UAV GNSS navigation data can be sufficient for many applications.