Local tropospheric model for conventional Repeat-Pass SAR interferometry

Abstract

Repeat-pass SAR Interferometry is a technique which produces an ‘interferogram’ giving earth surface deformation data. A major source of phase errors in interferogram is heterogeneous phase delay caused mainly by variation of water vapors in troposphere, which is a factor limiting applications of DInSAR mainly to desert or arid areas. This paper presented study results of three methods to correct atmospheric phase errors from water vapor in interferograms formed by TerraSAR-X images covered the south-eastern part of Bangkok metropolitan. The first method was the use of wet delay derived directly from MODIS precipitable water vapor product to correct the phase error in the interferograms. The second method employed ground-based meteorological data to calibrate MODIS PWV before computing phase delays. The third method improved the second method by estimating the expected MODIS PWV value at the time of the TerraSAR-X image acquisitions which over the Bangkok test area was 5 hour earlier than that of MODIS. The time-shifted linear fit model along with the IDW interpolation was used to estimate more realistic phase delays over entire imaging area. From the study, this time-corrected method provided best results while the second method also achieved a significantly better result than those obtained from direct use of MODIS PWV data. However, the accuracy of the calibration process should be further improved if GPS were integrated in the model and more GPS and ground meteorological stations are available.