Like other space geodetic techniques, interferometric synthetic aperture radar (InSAR) is limited by the variations of tropospheric delay noise. In this paper, we analyze the double-difference (DD) feature of tropospheric delay noise in SAR interferogram. By processing the ERS-2 radar pair, we find some tropospheric delay fringes, which have similar patterns with the GMS-5 visible-channel images acquired at almost the same epoch. Thirty-five continuous GPS (CGPS) stations are distributed in the radar scene. We analyze the GPS data by GIPSY-OASIS (II) software and extract the wet zenith delay (WZD) parameters at each station at the same epoch with the master and the slave image, respectively. A cosine mapping function is applied to transform the WZD to wet slant delay (WSD) in line-of-sight direction. Based on the DD WSD parameters, we establish a two-dimensional (2D) semi-variogram model, with the parameters 35.2, 3.6 and 0.88. Then we predict the DD WSD parameters by the kriging algorithm for each pixel of the interferogram, and subtract it from the unwrapped phase. Comparisons between CGPS and InSAR range changes in LOS direction show that the root of mean squares (RMS) decreased from 1.33 cm before correction to 0.87 cm after correction. From the result, we can conclude that GPS WZD parameters can be effectively used to identify and mitigate the large-scale InSAR tropospheric delay noise if the spatial resolution of GPS stations is dense enough.
|Number of pages||10|
|Journal||Journal of Atmospheric and Solar-Terrestrial Physics|
|Publication status||Published - 2006 Mar|
- Double difference
- Semi-variogram model
- Wet slant delay
- Wet zenith delay