Usefulness of Adaptive Filtering for Improved Digital Elevation Model Generation

The Digital Elevation Models (DEMs), which represent the variation of elevation in a terrain at spatial level, are an important source of input to a variety of applications for deriving a number of terrain parameters such as relative relief, slope, aspect direction etc. In recent years, Synthetic Aperture Radar Interferometry has been viewed as a powerful approach to derive quality DEMs from a pair of SAR images. Despite the interferometric technique is often limited by several de-correlations several researchers demonstrate its effectiveness in topographic mapping. The DEM accuracy is strongly influenced by the effectiveness of the phase unwrapping technique. In this study an effective adaptive filtering approach has been used to reduce the phase noise due to de-correlation and in improving the accuracy of phase unwrapping. Two well known phase unwrapping approaches such as branch cut and minimum cost flow network have been used. Interferometric data from ASAR sensor onboard ENVISAT satellite have been used. A highly undulated terrain condition near Dehradun city situated in Uttarakhand state of India was selected to investigate the performance of this adaptive filtering approach. The RMS error between the InSAR derived elevations and the map derived elevations was obtained as 7.2 m using adaptive filter. However, elevation map of the study area could not be generated due to high de-correlation effect without the use of adaptive filter. This result clearly demonstrates the effectiveness of adaptive filtering approach for generation of DEM at meter level accuracy, which is sufficient for many engineering applications.