Absolute geolocation of SAR Big-Data: The first step for operational InSAR time-series analysis

Abstract:

The time-series analysis of InSAR Big-Data generally involves three main steps including 1) a precise alignment of the stack of SAR images known as coregistration, 2) time-series analysis of a network of interferograms to obtain interferometric phase series, and 3) correction of the InSAR time-series to isolate the surface displacement signal. The alignment of the stack of SAR data can be achieved with a pure geometrical approach or with geometrical coregistration refined for mis-registrations derived from cross-correlation of SAR data. Since simplicity of workflows are generally preferred in an operational data system designed for Big-Data, a pure geometrical coregistration approach is favorable. Moreover, updating an existing coregistered stack with new acquisitions is simpler with a geometry-based approach. However, in practice the geometrical coregistration may be affected by signal propagation delay through ionosphere and troposphere or by solid earth tides. In this presentation, we evaluate the impact of each of these sources of mis-registration on stacks of C- and L-band data. The results demonstrate that the mis-registrations are dominated by the ionosphere and can be potentially predicted with GNSS-based Total-Electron-Content (TEC) measurements from receivers on the ground and on-board SAR platforms. We evaluate the expected performance of geometrical alignment refined with predicted mis-registrations of stacks of SAR data from the Sentinel-1 and NISAR missions.