Increasing Insar Coverage in Vegetated and Rough Terrain Using Temporal Stable Pixels

PS-InSAR has been proven to be a valuable technique for retrieving high accuracy crustal deformations by handpicking pixels with high phase stability identified as permanent scatterers. To increase the coverage area, pixels with stable spatial coherence throughout the InSAR stack have been used, which are identified as temporal coherent pixels (TCP). This technique excludes pixels with stable phase behavior because of their temporal varying neighboring pixels. In this paper, we identify the temporal correlation threshold of multi-looked pixels that maximizes analysis area coverage and ensures high phase stability using simulation analysis that relates spatial coherence with temporal phase correlation. The proposed methodology is used to retrieve crustal deformations of the 2011 Tohoku earthquake in Sendai region, Japan which contains rough and vegetated areas. The coverage area has been increased by 39% compared to TCP selection criteria and the retrieved deformations shows a RMSE of few millimeters in comparison to GPS observations.