DlnSAR method applied to dual-pair interferograms with Sentinel-1 data: a study case on inconsistent unwrapping outputs

摘要

Many works have been implemented to describe how seismic vertical displacement maps can be generated from synthetic aperture radar (SAR) images. The theory that supports these works is a very interesting field in remote sensing which is called differential interferometric SAR (DInSAR). However, behind this theory, there are many limitations and ambiguities that should be clarified before concluding specific results from the interpretation of subsidence maps. In this work, some of these drawbacks will be discussed, by following a typical subsidence map reconstruction with Sentinel-1 satellite images of a seismic region in Mexico. Particularly, the correct estimation of the unwrapped phase, is a crucial step for future analysis, like in time-series studies on subsidence ratios. In order to do this, a sophisticated routine called statistical-cost network-flow phase-unwrapping algorithm (SNAPHU), provided by the European Space Agency (ESA) in free software tools, is commonly applied to general SAR interferograms. However, in this particular study case, focused on simple dual-pair interferometry, atypical outputs obtained from SNAPHU were observed. Due to this inconsistency, the alternate solution proposed here is based on the so-called phase unwrapping by an accumulation of residual maps (ARM) algorithm. Our comparative analysis suggests that ARM outperforms SNAPHU and other tested methods in terms of unwrapped estimates.