InSAR points’ geolocation uncertainty estimation and geolocation improvement using LiDAR

Time Series Interferometric Synthetic Aperture Radar (TS-InSAR) technique can monitor displacement processes intarget areas with millimeter precision. However, limited by decimeter- or meter- level InSAR geolocation accuracy, directly associating InSAR points with actual ground targets merely based on InSAR-derived geolocation estimates is not always reliable. Especially for linear infrastructure like dams, poor-quality geolocation estimations can lead to deviations in the three- dimensional (3D) position of InSAR points, thereby failing to accurately link InSAR points with the specific structures of the dam. Here, we propose a method for 3D geolocation improvement of InSAR points based on the 3D error ellipsoid of InSAR positioning estimation and aided by Light Detection and Ranging (LiDAR) data (0.5 x 0.5 m). By establishing an error ellipsoid of every InSAR point and utilizing rotation and projection matrices for LiDAR datum transformation, we extract all LiDAR points located within the error ellipsoid and update InSAR point geolocation based on the extracted LiDAR values and their statistics. This process recalculates the 3-D geolocation of InSAR points and improves its accuracy. We applied this method to the Houtribdijk dam in the Netherlands, andimproved the InSAR points obtained with 152 and 148 Sentinel-1A IW SAR data (10 x 5m) in ascending and descending orbits acquired between 2018 and 2022, using the Actueel Hoogtebestand Nederland 3 (AHN3) LiDAR point cloud with centimeter-level accuracy. The results show that the InSAR points with 3D error ellipsoid properly link with structures over the entire dam compared with the points without concerning positioning uncertainty. For the SAR data in ascending and descending orbit, the Root Mean Square Errors (RMSE) of the heights between the LiDAR-based improved InSAR points and AHN3 LiDAR points are 0.4 m and 0.5 m, respectively. In contrast, the RMSE values for the InSAR points without LiDAR-based improvement are 1.4 m and 1.6 m, respectively. Furthermore, we compared the correlation of heights between all InSAR points on the dam and the AHN3-derived digital terrain model (DTM). The correlation of heights between the InSAR points without and with geolocation improvement and the AHN3 DTM is 0.14 and 0.95 with the RMSE values of 1.9 m and 0.5 m for ascending, 0.11 and 0.95 with the RMSE values of 1.2 m and 0.5 m for descending, respectively. All this demonstrates the efficacy of our method, and allows us to further precisely identify InSAR points from the slopes and top of the dam for the dam structures’ displacement assessment.   [1] Dheenathayalan P, Small D, Schubert A, et al. High-precision positioning of radar scatterers. Journal of Geodesy, 2016, 90(5): 403-422. [2] Chang L, Sakpal N P, Elberink S O, et al. Railway infrastructure classification and instability identification using Sentinel-1 SAR and laser scanning data. Sensors, 2020, 20(24): 7108. [3] Zhang, B., Chang, L., Stein, A., 2021. Spatio-temporal linking of multiple SAR satellite data from medium and high resolution Radarsat-2 images. ISPRS Journal of Photogrammetry and Remote Sensing, 176.