Deriving Urban Topography from Multi-baseline Sar Interferometric Phase Coherence Images

In this paper a method for deriving urban topography from multi-baseline SAR (synthetic aperture radar) interferometric phase coherence images is presented. This work is based on the premise that for flat (smooth) surfaces, coherence is retained even for large spatial baselines when the common band filter is applied. However, for rough terrain and elevated features this does not hold and coherence diminishes as baseline increases. Thus for areas containing large buildings, coherence drops off more rapidly with increasing baseline than for non-vegetated smooth surfaces (e.g. car parks). In this study, 87 co-registered coherence images of the cities of Cardiff and Bristol, UK were generated from subsets of 22 ERS scenes. The mean coherence for small windows covering the city, corresponding to different land covers, was calculated for all coherence pairs. By fitting theoretically derived curves to plots of mean coherence versus spatial baseline for a given stable area (window) and retrieving the ‘gradient’ of the curve it is possible to obtain a measure of urban surface roughness. Results show that for both Cardiff and Bristol the city centre exhibits the characteristics of a rougher surface while suburban residential areas are represented as smoother surfaces. This is to be expected since the city centre contains a high concentration of large buildings. Hence it is possible to differentiate between high-rise and low-rise districts using multiple coherence images in this way.