Non-parametric insar time series analysis of arcs using complex b-splines

Parametric models are widely utilized for interferometric synthetic aperture radar (InSAR) time series analysis under the assumption that the parameterization is invariant over time. Yet, the complexity of InSAR scatterers makes the hypothesis of this consistent and uniform behavior less likely. Here, we propose a method for non-parametric time series analysis for arcs between point scatterers based on basic splines (B-splines), which has the potential of fitting the time series adequately due to its high flexibility. We implement B-spline modelling for the kinematic behavior of arcs, and we propose to apply it in the complex domain and optimize the B-spline settings by means of amplitude data. We find that B-splines in the complex domain show great potential for estimating InSAR time series behavior, due to its insensitivity to errors in the integer ambiguity estimation, and that multiple model solutions can be derived from different B-spline parameter combinations. Our approach does not aim to provide a solution of the InSAR non-uniqueness problem, but it emphasizes the contribution of smoothness constraints to limit the solution space.