Near Real-Time InSAR Deformation Time Series Estimation With Modified Kalman Filter and Sequential Least Squares

Abstract

The current and planned synthetic aperture radar (SAR) sensors mounted on satellite platforms will continue to operate over the coming years, providing unprecedented SAR data for monitoring wide-range surface deformations. The near real-time processing of SAR interferometry (InSAR) data for the retrieval of ground-deformation time series is urgently required in the current era of big data. The state-of-the-art Kalman filter (KF) and sequential least squares (SLS) algorithms have been proposed to update an InSAR-driven ground-deformation time series. As a contribution of this study, we customize the conventional KF and SLS for big InSAR data for near real-time processing. The development of an accurate prediction model for KF-based InSAR processing is a challenge owing to the large scale of the targets for surface monitoring. We developed a modified KF algorithm, abbreviated as npKF, that does not require any prediction information, abbreviated as npKF. In this context, to avoid occupying a large storage space in SLS-based InSAR processing, we developed a modified SLS algorithm with a truncated cofactor matrix, abbreviated as TSLS. Using both simulated and actual SAR data, we evaluated the performance of these methods under three different aspects: accuracy, computation, and storage performance. With big data, the proposed method can estimate the deformation time series in near real time. It will be a reliable and effective tool for producing near real-time InSAR deformation products in the coming era of processing big SAR data and will play a part in the geologic hazard routine monitoring and early warning system.

Keywords

• Big synthetic aperture radar interferometry (InSAR) data
• deformation time series
• Kalman filter (KF)
• sequential least squares (SLS)