Pixel-based interferometric pair selection in InSAR time-series analysis with baseline criteria

Standard synthetic aperture radar interferometry (InSAR) time-series analysis uses interferometric pairs selected using arbitrarily fixed baseline criteria, and pixels coherent within the baseline criteria are used for displacement estimations. However, adequate baseline criteria could differ pixel by pixel, especially in suburban areas, because coherence variation depends on the scattering characteristics of specific pixels. Accordingly, it would be better to determine the optimal baseline criteria with which to select interferometric pairs on a pixel-by-pixel basis. In this study, we developed such a method by satisfying a coherence threshold and by maximizing the accuracy of surface displacement velocity. This selection of optimal interferometric pairs enables us to use a greater number of interferometric pairs for highly coherent pixels, which can improve the accuracy of the estimated surface displacements. Furthermore, it enables the estimation of displacement with a minimal number of interferometric pairs for noisy pixels, which can increase the number of pixels available for mapping surface displacement. We examined our method by estimating surface displacement velocity around the Ohaaki geothermal field in New Zealand. The results showed an increase in the number of coherent pixels available to map surface displacement velocity. Accordingly, the area of surface displacement was defined more clearly compared with the standard analysis.