Unmanned aerial vehicles (UAVs) have been used to survey forests for decades. To monitor post-forest operation effects such as the surface deformation of forest soils (e.g., soil erosion and deposition), the alignment of 3D surface models using ground control points (GCPs) by georeferencing is necessary. However, the GCP network optimization for surveying forests has not been clearly established. We installed 29 GCPs on tree stumps in a post-timber harvesting site to spatially correct UAV survey results, for which real-time kinematics data were unavailable. The GCPs for georeferencing were randomly selected from 13 validation sets for each of the three trials in this study; the GCP and checkpoint root mean square errors (RMSEs) were used for determining the georeferencing results. Even if the number of GCPs increased, the GCP RMSE did not decrease; the checkpoint RMSE decreased by 2.03 cm. The 3D surface model alignment using 6 and 26 GCPs did not show a significant difference in alignment error (3.78 and 2.96 cm, respectively). Our results reveal that GCP RMSE does not affect the alignment error; therefore, we suggest using at least 6 GCPs for the precise alignment of 3D surface models while utilizing the same GCPs for aligning pre- and post-3D surface models.