Assessment of the Positional Accuracy of PPK-Based UAS Orthomosaics and DSM according to PEC-PCD over Steep Slopes

Abstract

This study compares the positional accuracy of photogrammetric products: orthomosaic and digital surface model (DSM), generated in a steep slope area in Recife, Brazil, using two UAS platforms: the DJI Mavic 3 Enterprise (M3E) with RTK/PPK and the DJI Mini 3 operating in PPK mode via the T2R solution, with processing performed in both open-source (WebODM/ODM) and proprietary (Agisoft Metashape) environments. The objective was to evaluate horizontal and vertical accuracy without the use of ground control points (GCPs) during processing, employing GCPs exclusively as checkpoints for the computation of positional discrepancies, horizontal, vertical, and three-dimensional RMS, as well as classification according to the Brazilian Cartographic Accuracy Standard for Digital Cartographic Products (PEC-PCD). The processing workflow included rolling shutter correction (for the DJI Mini 3), enforced use of GNSS-derived image coordinates (force-gps) with an adopted accuracy of 0.02 m for camera perspective centers (gps-accuracy), and PPK-corrected geotags obtained via GNSS processing in RTKLib and T2R-Geotagger. Flights were conducted at 100 m altitude (terrain following), with 85%/75% forward/lateral overlap and a speed of 6 m/s. The results, evaluated in QGIS/LFTools, showed that in Metashape the Mini 3 achieved RMS_XY ≈ 9.3 cm and RMS_Z ≈ 11.5 cm, while the M3E reached RMS_XY ≈ 5.5 cm and RMS_Z ≈ 4.5 cm. In WebODM, the Mini 3 achieved RMS_XY ≈ 13.1 cm and RMS_Z ≈ 18.2 cm, whereas the M3E obtained RMS_XY ≈ 6.4 cm and RMS_Z ≈ 4.8 cm. Regarding planimetric PEC-PCD classification, the Mini 3 exhibited equivalent performance in both environments — Class B at 1:500 scale and Class A from 1:1,000 onward — while the M3E achieved Class A at all evaluated scales. For vertical PEC-PCD classification, the Mini 3 showed better performance in Metashape (Class B at 1:500 and Class A at 1:1,000), whereas in WebODM a reduction in quality was observed (Class C at 1:500 and Class B at 1:1,000), and the M3E maintained Class A at all scales. It is concluded that the M3E ensures the highest positional accuracy regardless of processing software. However, the Mini 3 operating in PPK demonstrates the capability to produce products compliant with Brazilian cartographic accuracy standards, achieving PEC-PCD Class A up to 1:1,000 in planimetry and up to 1:2,000 in altimetry, representing a lower-cost, high-mobility solution suitable for slope monitoring, infrastructure assessment, and risk management.