ACCURACY ASSESSMENT OF THAILAND’S NETWORK REAL TIME KINEMATIC (NRTK) FOR UNMANNED AERIAL VEHICLE (UAV) PHOTOGRAMMETRY

Abstract

Unmanned aerial vehicle (UAV) photogrammetry is a technology used to create highly accurate maps, and the real-time kinematic (RTK) global navigation satellite system (GNSS) network has been implemented by several organizations to support their missions. Recent advancements in computer vision have driven progress in photogrammetry, alongside the development of more accessible UAVs equipped with GNSS technology. This paper evaluates the accuracy of UAV imagery and processing using ground control points (GCPs) with coordinates from the Thailand RTK GNSS network. The study area is located on the Sathing-Phra Peninsula in Songkhla Province, southern Thailand, covering approximately 174 hectares with a total of 660 images. The average ground sampling distance (GSD) was around 4.35 cm. The processed photogrammetry system used 8 GCPs via the GNSS RTK network, and results were compared with 25 checkpoints from a static survey. The UAV photogrammetry results showed a horizontal accuracy for orthomosaics and vertical accuracy for the digital surface model (DSM), with CE 95 values of 6.99 cm and 12.44 cm, respectively, as per the NSSDA standard. These results comply with the Thailand UAV Surveying for Engineering (TUSE) standards. According to the 2014 American Society for Photogrammetry and Remote Sensing (ASPRS) standards, horizontal accuracy class I and vertical accuracy class II should be applied. The map scale factor (MSF) standard was 1:160, with a contour interval of 20 cm. Additionally, the orthomosaic DSM can be imported into open-source programs for use in a web GIS online system, enabling local agencies to retrieve and utilize the data.