The Influence of Oblique Views and Ground Control Point Distribution on Unmanned Aerial Systems Photogrammetry Precision for Pavement Assessment

Over the last two decades, unmanned aerial systems (UAS) have gained increased traction as a tool for pavement inspection, particularly for evaluating rutting, apparent lift thickness, and surface irregularities. Despite their widespread use, a lack of consensus remains on the optimal number and distribution pattern of ground control points (GCPs) for accurate georeferencing. In this study, six test configurations were created using two types of imagery (nadir and oblique) and varying GCP patterns to assess their influence on the accuracy of digital elevation models over a 50 × 50 m pavement test area. A combination of geostatistical methods, including kriging and inverse distance weighting, along with metrics such as root mean square error, mean absolute error, and bias, was used to evaluate performance. The results show that oblique imagery significantly improves model accuracy when GCPs are limited. While increasing the number of GCPs beyond five did not yield notable improvements, error metrics and semivariogram behavior remained consistent across patterns. However, spatial analysis tools, including Getis-Ord Gi* and Moran’s I, revealed localized clusters of high error, particularly in configurations lacking oblique coverage. These findings underscore the importance of evaluating not only global accuracy measures, but also spatial patterns of error when designing UAS-based pavement surveys. The results support the conclusion that a minimum number of strategically distributed GCPs, in the order of approximately five per 2,500 m 2 , combined with oblique imagery, can provide reliable elevation modeling for small-scale pavement assessment.