Application of the Coherent Pixels Technique (CPT) to large datasets of ERS and ENVISAT SAR images for deformation time series retrieval

Jordi J. Mallorqui(1) , Pablo Blanco(1) , Jesus Sanz-Marcos(1) , Sergi Duque(1) , and David Navarrete(1)

(1) Universitat Politecnica de Catalunya (UPC), D3-Campus Nord-UPC, 08034 Barcelona, Spain

Abstract

In this paper, a modification of the Coherent Pixels Technique (CPT) to work with large datasets of both ERS and ENVISAT images will be presented. CPT is able to retrieve the linear and non-linear components of movement from a set of low resolution interferograms (multi-looked), estimating at the same time the DEM error and the atmospheric artefacts. The basis for the linear estimation of movement is the adjustment of a linear model, which considers the linear velocity of displacement and the DEM error, to the available data. The pixel selection criterion initially implemented was based on coherence stability in the stack of interferograms, leading to a lower resolution product and a limitation on the maximum useful spatial baselines. As the basic CPT approach is almost independent on the pixel selection criteria, an amplitude one based on the classical amplitude dispersion thresholding has also been included, which allows working at the maximum resolution of the SLC image. The performance of both approaches over the same testsite will be compared. Besides this, the generation of the interferograms does not require establishing a master image, allowing free combinations between all available images depending on the restrictions of temporal and perpendicular baselines. The selection of the best set of interferograms from all possible is indeed an interesting point. An “optimal interferogram selection” has been implemented in order to minimize the existence of subsets and maximize the distribution of baselines, both spatial and temporal, to improve final results. The impact of such selection on the final results will also be addressed in the paper. The method adjusts a linear model to phase increments between two neighbouring pixels linked with the Delaunay triangulation, avoiding the need of a sparse grid phase unwrapping of the interferograms. Once the linear velocity of deformation and the DEM error have been retrieved, the algorithm continues with the non-linear movement and the atmospheric artefacts estimation. One of the advantages of the algorithm is that there is no need to unwrap the noisy differential interferograms, which can be a difficult step and a potential source of errors, but this can be at the same time a limitation when strong non-linearities are present on the deformation evolution. The need of sparse-grid phase unwrapping will be evaluated. Finally, results with real data will be presented using the test-sites of Gardanne (F), Paris (F) and Murcia (S).

This work has been financed by ESA under projects “CPT: Development of algorithms for the exploitation of ERS-ENVISAT using the Coherent Pixels Technique” (Contract 17584/03/I-LG) and “PSIC4 (Persistent Scatters Interferometry): Processing Over a Validation Test Site” (Contract 18708/04/I-LG).

 

Full paper

 

  Higher level                 Last modified: 07.10.03