Comparing terrestrial laser scanner and ground-based structure from motion photogrammetry for urban sinkhole characterization and monitoring in Zaragoza, Spain

Abstract

Delimiting, characterizing and monitoring active sinkholes in urban areas are fundamental steps for effectively managing the associated risks. These high-exposure scenarios require accurate data on hazard parameters (e.g., spatially distributed subsidence rates), but the current investigation and monitoring techniques for sinkholes remain relatively undeveloped in comparison to other geological hazards, such as landslides. In this regard, we present the first comparative analysis of the performance of terrestrial laser scanner (TLS) and ground-based structure from motion (SfM) photogrammetry for delimiting the actively deforming areas and characterizing the spatial patterns of ground displacement through the comparison of two pairs of high-resolution 3D point clouds. To assess their performance, this work utilizes vertical displacement data measured by high-precision levelling. The main finding is that, despite TLS providing displacement data with less noise and internal distortion, the less expensive and easier-to-implement SfM photogrammetry using ground-based conventional cameras yields a comparable performance when accurate geodetic data is available. However, according to high-precision levelling data, both techniques may underestimate the extent and rate of the deformation. According to levelling data, the active sinkhole has a major axis 60 m long, while the length detected by TLS and SfM photogrammetry drops to 21 and 17 m, respectively. Maximum subsidence rates by levelling, TLS and SfM were 22.9, 17.2 and 15.2 mm/year, respectively. These results indicate that there is still a need to complement these high-resolution techniques with the use of high-precision methods such as levelling or additional geodetic benchmarks.