The Influence of Ground Control Points (GCPs) for Digital Elevation Models (DEMs) Using UAVs

Abstract

Unmanned aerial vehicles are a modern technology for collecting data for aerial surveying to produce high-resolution digital elevation models (DEMs). This accuracy is achieved using ground control points (GCPs), which are specific locations on the ground with known coordinates that help improve the precision of aerial data collection. Digital elevation models are used in many applications such as topographic mapping, land-use planning, and levelling. This study aims to identify the optimum number of GCPs to achieve high accuracy for digital elevation models and reduced fieldwork. 1 km2 was imaged with a commercial UAV named Matrice 300 at a height of 200 m above the ground, with 80% forward overlap and 70% side overlap. 25 ground control points and 5 check points were created using the GPS technique. The images were processed using Agisoft Metashape software by an adjustment process with an incremental number of GCPs starting with 4 and ending with 25. The remaining points were used as a checkpoint to determine the precision of the model at each trial. Root Mean Square Error (RMSE) was used as an accuracy measurement. The study demonstrates that horizontal accuracy (RMSE XY) stabilizes after 5 ground control points (GCPs), with a marginal enhancement from 1.2 cm to 1.1 cm after 13 GCPs. The vertical accuracy (RMSE Z) improves markedly from 3.1 cm to 2.1 cm with 18 GCPs, demonstrating that height precision is more responsive to the quantity of control points. The 3D accuracy (RMSE XYZ) enhances by 1 cm with the addition of ground control points (GCPs) up to 18 but becomes stable at 2.3 cm after that, which is consistent with the vertical accuracy measurements. A medium GCP density of 9–12 is recommended for general UAV-based digital elevation model (DEM) generation, balancing accuracy and efficiency.