Kinematic Coregistration of Sentinel-1 TOPSAR Images Based on Sequential Least Squares Adjustment

• Main Authors: Bing Xu, Zhiwei Li, Yan Zhu, Jiancun Shi, Guangcai Feng
• Format: Article
• Language: English
• Published: IEEE 2020-01-01
• Series: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
• Subjects: Least squares adjustment; multitemporal terrain observation by progressive scans mode Synthetic Aperture Radar (TOPSAR) images coregistration; Sentinel-1; sequential weighted least square adjustment (SWLSA)
• Online Access: https://ieeexplore.ieee.org/document/9108545/

The Sentinel-1 provides an unprecedented opportunity for InSAR research and applications, especially in the field of fast and accurate damage assessment, thanks to its extra wide swath, short revisit interval, and free policy. Challenges also exist in Sentinel-1 terrain observation by progressive scans mode synthetic aperture radar (TOPSAR) interferometric processing, for example, the coregistration of TOPSAR images requires an accuracy of 0.001 pixels to reduce the phase jumps at the burst overlap region to 3°. To obtain the accuracy of 0.001 pixels for the coregistration of a stack of multitemporal TOPSAR images, joint estimation method and network-based method were proposed and implemented statically. However, when new images are added, the existing methods cannot coregister them kinematically. In order to resolve this issue, we first give a brief review for the existing static methods, including the single master-, temporally transferred-, and network-based methods, for coregistering multitemporal TOPSAR images. Then, we propose a kinematic coregistration method to coregister newly added TOPSAR images by introducing the sequential weighted least square adjustment. Experimental results demonstrate that the proposed method can achieve an accuracy of 0.001 pixels for kinematic coregistrations of multitemporal TOPSAR images. Compared with the static network-based coregistration method, the proposed method is superior in terms of both coregistration accuracy and computational efficiency. It will contribute a great deal to the globally acquired big SAR data (e.g., Sentinel-1 TOPSAR) and their near real-time processing.