Coherence-based Methodology for Interferometric Dem Integration
Pablo Euillades(1) , Leonardo Euillades(1)
, Mauro Blanco(1)
, and Gabriel Cabrera(1)
Instituto CEDIAC - Universidad Nacional de Cuyo,
Ciudad Universitaria - Parque Gral San Martin,
High resolution-interferometric digital elevation models often present artifacts or lack of coverage over low coherence areas. These effects arise when the area of interest (AOI) is mountainous or flooded, even when using a priori optimal -in temporal and spatial baseline sense- pairs. If the AOI is located in a country where only a few SAR data has been acquired, the possibilities to obtain a good DEM are even worse.
In this work, we propose a methodology to integrate two or more interferometric DEMs. The key point is using images acquired under different conditions, i.e. different dates (useful for areas susceptible of floods) or different acquisition geometry (for mountainous areas). The algorithm takes the coregistered coherence maps of the original DEMs and, after a filtering step, calculates a Join Coherence Map (JCM). The JCM point at the better original DEM in a pixel by pixel basis, thus allowing the extraction of complementary coverage from them. The last step is to remove the remaining discontinuities, which is achieved using an deformation algorithm based in a piecewise linear interpolation over a dense grid of control points.
To illustrate the methodology we present a real case study located in a mountainous AOI of western Argentina, where only a few ERS scenes exists. Here, one ascending-pass and one descending-pass DEMs have been integrated. Each one of them alone present many holes without data and unwrapping artifacts. When fused, the result is a much better DEM, covering most of the AOI.