UAV-Based Slope Mapping Accuracy at Different Flight Altitudes Using Minimum Ground Control Point Configurations

Abstract

This study evaluates the accuracy of Unmanned Aerial Vehicle (UAV)-based slope mapping at different flight altitudes using minimal Ground Control Point (GCP) configurations. Two 3-GCP arrangements, namely diagonal top-to-bottom and straight top-to-bottom, were first compared using Real-Time Kinematic (RTK)-based cross-sectional validation. The more reliable diagonal configuration was then evaluated at 50, 75, and 100 m flight altitudes using GCP-based accuracy metrics, RTK profile comparison, CloudCompare-based Multiscale Model to Model Cloud Comparison (M3C2) surface-deviation analysis, and Cumulative Distribution Function (CDF) analysis. The results show that the diagonal arrangement more consistently reproduced slope geometry than the straight configuration. Under the diagonal configuration, the 50 m dataset produced the largest GCP residuals, profile errors, and surface deviations despite having the finest Ground Sampling Distance (GSD). In contrast, the 75 m and 100 m datasets exhibited significantly lower deviations and demonstrated generally comparable performance. The 75 m dataset yielded the lowest total GCP error (28.73 mm) and slightly lower profile and percentile-based deviation metrics, whereas the 100 m dataset produced slightly lower average M3C2 surface-deviation statistics. These findings indicate that a finer GSD and a lower flight altitude do not necessarily result in higher slope-mapping accuracy. Under the tested slope, UAV platform, and minimum-GCP condition, the diagonal configuration provided better geometric reconstruction than the straight configuration, while the 75 m and 100 m flight altitudes produced comparable and more reliable results than the 50 m flight.