Terrestrial and Airborne Structure from Motion Photogrammetry Applied for Change Detection within a Sinkhole in Thuringia, Germany

Abstract

Detection of geomorphological changes based on structure from motion (SfM) photogrammetry is highly dependent on the quality of the 3D reconstruction from high-quality images and the correspondingly derived point precision estimates. For long-term monitoring, it is interesting to know if the resulting 3D point clouds and derived detectable changes over the years are comparable, even though different sensors and data collection methods were applied. Analyzing this, we took images of a sinkhole terrestrially with a Nikon D3000 and aerially with a DJI drone camera in 2017, 2018, and 2019 and computed 3D point clouds and precision maps using Agisoft PhotoScan and the SfM_Georef software. Applying the “multiscale model to model cloud comparison using precision maps” plugin (M3C2-PM) in CloudCompare, we analyzed the differences between the point clouds arising from the different sensors and data collection methods per year. Additionally, we were interested if the patterns of detectable change over the years were comparable between the data collection methods. Overall, we found that the spatial pattern of detectable changes of the sinkhole walls were generally similar between the aerial and terrestrial surveys, which were performed using different sensors and camera locations. Although the terrestrial data collection was easier to perform, there were often challenges due to terrain and vegetation around the sinkhole to safely acquire adequate viewing angles to cover the entire sinkhole, which the aerial survey was able to overcome. The local levels of detection were also considerably lower for point clouds resulting from aerial surveys, likely due to the ability to obtain closer-range imagery within the sinkhole.

Appraisal of Ancient Quarries and WWII Air Raids as Factors of Subsidence in Rome: A Geomatic Approach

“…Considering also the achievements reached in the study of analogous cases in Italy and out of the country, further efforts have been directed toward identifying precursory factors [29][30][31][32]. Geomatics monitoring by space-borne, close-range and ground-based platforms bearing a wide range of sensors has been validated for the evidencing of soil raveling, using, as an example, photogrammetric modeling [33,34], ground-based systems [35], geophysics approaches [17,21,26,[36][37][38][39], or sensor networks [40]. The Differential Interferometric Synthetic Aperture Radar (DInSAR) technique, one of the key methods applied in satellite monitoring in urban areas [41][42][43][44], has also recently been applied to the investigation of sinkholes ( [29,31,32,[45][46][47], and a preliminary approach in [48]).…”