Abstract-Reconstructing high-accuracy Digital Elevation Model (DEM) is influenced by phase errors, such as phase trend, low coherence problems and phase unwrapping. These problems could result in the conversion errors from the phase to height. In this paper, a method is proposed to reconstruct the high-accuracy DEM using satellite interferometric synthetic aperture radar (InSAR). The proposed algorithm mainly aims to reduce the phase errors from the phase trend and low coherence problems. It consists of three steps. Firstly, the orbit state vectors are precisely interpolated in 3-D coordinates rather than in a separate dimension with the exploration of the orbital elements. Secondly, the relationship between external DEM and the interfermetric phase is built by the improved precise geo-location algorithm. The phase trend is estimated according to the topographic information and then removed from the unwrapped interferogram. Thirdly, the interferogram in low coherent regions are all updated with the simulated phases from actual DEM. The accuracy of the InSAR derived DEM can be significantly improved without any
16Bin, Xia, and Zheng ground control points (GCPs), especially in those regions contaminated by masses of residues. Meanwhile, the phase trend caused by atmosphere effects or orbits uncertainty can also be eliminated by using this method. The experiment has demonstrated the proposed method can yield quite satisfactory results for producing high-accuracy DEM using Envisat data.