On the use of VR for improved interpretation of InSAR results

Abstract

Understanding surface deformation is critical for monitoring and maintaining urban structures and environments. Persistent Scatterer Interferometry (PSI) plays an important role in this, it can give insight regarding the stability and deformation of both anthropogenic and natural structures. Despite its utility, the interpretation is a difficult process, representing a four-dimensional problem with multiple sources of uncertainty. Current two-dimensional visualization methods are lacking in conveying all information inherent to PSI data, which lead to the aim of this research, which was to develop and design a prototype Virtual Reality (VR) application that can enhance the interpretation of PSI data by integrating the visualization of PSI data with contextual 3D geometric data. A Unity based VR prototype was developed that visualizes sub-pixel and pixel-level PSI datasets alongside LiDAR point clouds, 3D Models, Building Information Models (BIM) and 3D geometric properties inherent to InSAR. Three diverse case studies, a subsiding railway (Betuwelijn), a collapsed tower surrounded by forests and hills (Wilhelminatower) and a suburban environment (TU Delft Campus), were selected to visualize in VR. In the application users can navigate freely, select individual scatterers, visualize their $2\sigma$ confidence intervals, their geometric properties and view time-series in situ in an environment visualized by the 3D geometric datasets. Six domain experts evaluated the system through open exploration and structured interviews. Results showed that the enhanced depth perception and interactive environment improved the user's ability to link deformation signals to real-world features, identify anomalies and assess uncertainty. Despite it's current usability limitations and limited functionality, VR proved a powerful platform for multi-dimensional data exploration, offering new opportunities for monitoring infrastructure. It was concluded that through the integration of geometric datasets, such as LiDAR, 3D models and geometric properties, the interpretation of PSI data is enhanced by accessing the data in a higher dimension and linking it to the environment.