A Hybrid Unwrapping Error Correction Method based on Phase Closure and Bridging for a Network of Interferograms
Abstract
InSAR time-series estimated from the inversion of a network of interferograms can be potentially biased due to the wrong phase jump added to the interferometric phase during phase unwrapping, called unwrapping errors. Despite of the existence of several three-dimensional phase unwrapping algorithms, the two-dimensional phase unwrapping algorithms seem to be more popular and commonly used in the InSAR processing because of its simplicity and efficiency of algorithm such as SNAPHU (Chen and Zebker, 2001). Thus, it's relevant to develop methods to evaluate the interferometric phase unwrapped by two-dimensional algorithms and possibly detect and correct the unwrapping error.
Here we proposed a hybrid method to automatically detect and correct the unwrapping errors in the network of interferograms by combining the phase closure constraint in time domain (Fattahi, 2015) and bridging of connected components in space domain. More specifically, we detect the unwrapping errors based on the non-zero phase closure (multiple of 2 pi component) of all possible phase triplets, identify and label the unwrapping errors occurred within the coherent connected components which is correctly unwrapped internally (Fig. 1a), then correct it using bridging method based on the phase spatial continuity. The proposed method does not require a dense redundant network with majority of them unwrapping error free. It's suitable for correcting unwrapping error occurred on areas separated by narrow decorrelated features such as narrow water bodies or steep topography. We tested the proposed method with Sentinel-1 dataset on Galapagos Islands where both Fernandina and Santiago islands suffer from unwrapping errors due to water separation. We compare the temporal coherence computed from the inversion of the interferometric phase before the correction, after correction with phase closure (Fattahi, 2015) and after correction with proposed method (Fig. 1). The temporal coherence on Fernandina increased from 0.56 to 0.91, to 0.99 while the temporal coherence on Santiago increased from 0.05 to 0.30, to 0.76. The final displacement time-series and Line-Of-Sight (LOS) linear velocity are successfully estimated on both islands (Fig. 2), together the co-eruptive deformation on the 2017 Sep eruption on Fernandina volcano (dashed line on Fig. 2c).- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.G41B0690Z
- Keywords:
-
- 1240 Satellite geodesy: results;
- GEODESY AND GRAVITYDE: 1241 Satellite geodesy: technical issues;
- GEODESY AND GRAVITYDE: 1908 Cyberinfrastructure;
- INFORMATICSDE: 1932 High-performance computing;
- INFORMATICS