Georeferencing using ground control points (GCPs) is the most common strategy in photogrammetry modeling using unmanned aerial vehicle (UAV)-acquired imagery. With the increased availability of UAVs with onboard global navigation satellite system–real-time kinematic (GNSS RTK), georeferencing without GCPs is becoming a promising alternative. However, systematic elevation error remains a problem with this technique. We aimed to analyze the reasons for this systematic error and propose strategies for its elimination. Multiple flights differing in the flight altitude and image acquisition axis were performed at two real-world sites. A flight height of 100 m with a vertical (nadiral) image acquisition axis was considered primary, supplemented with flight altitudes of 75 m and 125 m with a vertical image acquisition axis and two flights at 100 m with oblique image acquisition axes (30° and 15°). Each of these flights was performed twice to produce a full double grid. Models were reconstructed from individual flights and their combinations. The elevation error from individual flights or even combinations yielded systematic elevation errors of up to several decimeters. This error was linearly dependent on the deviation of the focal length from the reference value. A combination of two flights at the same altitude (with nadiral and oblique image acquisition) was capable of reducing the systematic elevation error to less than 0.03 m. This study is the first to demonstrate the linear dependence between the systematic elevation error of the models based only on the onboard GNSS RTK data and the deviation in the determined internal orientation parameters (focal length). In addition, we have shown that a combination of two flights with different image acquisition axes can eliminate this systematic error even in real-world conditions and that georeferencing without GCPs is, therefore, a feasible alternative to the use of GCPs.

Using a GNSS RTK (Global Navigation Satellite System Real Time Kinematic) -equipped unmanned aerial vehicle (UAV) could greatly simplify the construction of highly accurate digital models through SfM (Structure from Motion) photogrammetry, possibly even avoiding the need for ground control points (GCPs). As previous studies on this topic were mostly performed using fixed-wing UAVs, this study aimed to investigate the results achievable by a quadrocopter (DJI Phantom 4 RTK). Three image acquisition flights were performed for two sites of a different character (urban and rural) along with three calculation variants for each flight: georeferencing using ground-surveyed GCPs only, onboard GNSS RTK only, and a combination thereof. The combined and GNSS RTK methods provided the best results (at the expected level of accuracy of 1–2 GSD (Ground Sample Distance)) for both the vertical and horizontal components. The horizontal positioning was also accurate when georeferencing directly based on the onboard GNSS RTK; the vertical component, however, can be (especially where the terrain is difficult for SfM evaluation) burdened with relatively high systematic errors. This problem was caused by the incorrect identification of the interior orientation parameters calculated, as is customary for non-metric cameras, together with bundle adjustment. This problem could be resolved by using a small number of GCPs (at least one) or quality camera pre-calibration.

Lately, affordable unmanned aerial vehicle (UAV)-lidar systems have started to appear on the market, highlighting the need for methods facilitating proper verification of their accuracy. However, the dense point cloud produced by such systems makes the identification of individual points that could be used as reference points difficult. In this paper, we propose such a method utilizing accurately georeferenced targets covered with high-reflectivity foil, which can be easily extracted from the cloud; their centers can be determined and used for the calculation of the systematic shift of the lidar point cloud. Subsequently, the lidar point cloud is cleaned of such systematic shift and compared with a dense SfM point cloud, thus yielding the residual accuracy. We successfully applied this method to the evaluation of an affordable DJI ZENMUSE L1 scanner mounted on the UAV DJI Matrice 300 and found that the accuracies of this system (3.5 cm in all directions after removal of the global georeferencing error) are better than manufacturer-declared values (10/5 cm horizontal/vertical). However, evaluation of the color information revealed a relatively high (approx. 0.2 m) systematic shift.

Most ground filtering algorithms are primarily designed for airborne LiDAR point cloud processing and their successful use in identifying ground points from photogrammetric point clouds remains questionable. We compared six ground filtering algorithms implemented in Metashape, ArcGIS, CloudCompare, LAStools, and PDAL. We used UAV photogrammetry-based (acquired under leaf-off conditions) and airborne LiDAR (leaf-on) point clouds of the same area to: (i) compare accuracy of generated DTMs; (ii) evaluate the effect of vegetation density and terrain slope on filtering accuracy; and (iii) assess which algorithm parameters have the greatest effect on the filtering accuracy. Our results show that the performance of filtering algorithms was affected by the point cloud type, terrain slope and vegetation cover. The results were generally better for LiDAR (RMSE 0.13–0.19 m) than for photogrammetric (RMSE 0.19–0.23 m) point clouds. The behavior in varying vegetation and terrain conditions was consistent for LiDAR point clouds. However, when applied on photogrammetric point clouds, the algorithms’ behavior was inconsistent, especially in areas of steep slope (except for the Progressive Triangulated Irregular Network in LAStools). Parameters related to the selection of the initial minimum elevation ground points were the most influential in all algorithms and point clouds.

In this paper, we evaluated the results in terms of accuracy and coverage of the LiDAR-UAV system DJI Zenmuse L1 and Digital Aerial Photogrammetric system (DAP – UAV) DJI Zenmuse P1 in a forested area under leaf-off conditions on three sites with varying terrain ruggedness/tree type combinations. Detailed reference clouds were obtained using terrestrial scanning by Leica P40. Our results show that branches pose no problem to the accuracy of LiDAR-UAV and DAP-UAV derived terrain clouds. Elevation accuracies for photogrammetric data were even better than for LiDAR data – as low as 0.015 m on all sites. However, the LiDAR system provided better coverage, with almost full coverage at all sites, while the DAP-UAV coverage declined with the increasing density of branches (being worst in the young forest). In the very dense young forest (Site 1), the coverage by photogrammetrically extracted terrain cloud using high calculation quality and no filtering achieved 80.7% coverage, while LiDAR-UAV reached almost 100% coverage. The importance of the use of multiple (or last) returns when using LiDAR-UAV systems was demonstrated by the fact that on the site with the densest vegetation, only 11% of the ground points were represented by first returns.

The current trend in the use of remote sensing technologies is their use as a tool for monitoring hard-to-reach areas, objects or phenomena in the alpine environment. Remote sensing technology is also effectively used to monitor geohazards and the development of human-made changes in the country. Research presented in this study demonstrates the results for the usability of the publicly available national digital elevation model DEM 5.0 obtained by utilizing the airborne laser scanning (ALS) survey to monitor the development of erosion, morphological changes of talus cones, or the dynamics of movement of rock blocks between stages of measurement in the alpine environment of the High Tatras mountains. The reference methods for this study are the terrestrial laser scanning (TLS) and structure-from-motion (SfM) photogrammetric approach using unmanned aerial systems (UASs). By comparing the created DEMs, the ALS point cloud’s accuracy on mostly rocky areas of different sizes was verified. The results show that the standard deviation of the ALS point cloud ranges from 19 to 46 mm depending on the area’s size and characteristics. The maximum difference ranges from 100 to 741 mm. The value of systematic displacement of data obtained by different technologies ranges from 1 to 29 mm. This research confirms the suitability of the ALS method with its advantages and limits for the detection of movement of rock blocks or change of position of any natural or anthropogenic objects with a size from approximately 1 m².