Multi-dimensional Data Tensor-based Modeling of GNSS-based InSAR for Hourly 3D Deformation Retrieval

Global Navigation Satellite System based interferometric synthetic aperture radar (GNSS-based-InSAR) can overcome the limitations of repeat-pass time by receiving signals from different navigation satellite combinations at different times, enabling hourly three-dimensional(3D) deformation monitoring. However, this monitoring mode requires joint processing of scene reflected signals from multiple satellites in multiple time periods, resulting in massive data volumes. Traditional InSAR signal models, which only consider single-satellite repeat-pass scenarios, fail to capture the relationships among time, space, and satellite dimensions in GNSS-based-InSAR system data, posing challenges for high-precision 3D deformation field retrieval. In this paper, a tensor-based multi-dimensional model is proposed for GNSS-based InSAR. First, the projection relationships of short-interval, multi-satellite GNSS-based InSAR data in the time, space, and satellite dimensions are established, and the relationships and physical significance of the data across different dimensions and profiles are analyzed. Second, based on the proposed tensor model, PS selection, interferometry, and hourly 3D deformation retrieval are implemented. The tensor model enables real-time analysis of data relationships across different dimensions, facilitating the development of error compensation algorithms and improving accuracy. Finally, using BeiDou system data, hourly accumulative 3D deformation field retrieval is achieved, with accuracy 4.68mm, 4.87mm, 7.34mm along East-West, North-South, Up-Down direction, demonstrating the effectiveness of the proposed tensor-based model for GNSS-based InSAR.