Baseline recognition and parameter estimation of persistent-scatterer network in radar interferometry

Similar to GPS sites, persistent scatterers (PSs) identified from a time series of radar interferograms can be used to establish a network for monitoring long-term ground deformation. We propose an adjacent array model for searching PS to PS connection (baseline) to form a Delaunay triangular network. The algorithm of temporal coherence maximization is employed to estimate the increments of deformation velocities and elevation errors along each PS-PS connection. The baseline recognition and parameter estimation methods are applied to detect land subsidence in Hong Kong. The algorithm validation is performed using SAR images collected over Hong Kong by the ASAR sensor onboard satellite Envisat during 2006 - 2007. The GPS measurements at 12 sites are used to correct atmospheric effects in the interferograms and calibrate the PS solution. Test results show that the proposed methods are viable and reliable for detecting ground deformation. The achievable accuracy of linear deformation velocity is about ±2.0 mm/a.