High-quality three-dimensional displacement fields from new-generation SAR imagery: application to the 2017 Ezgeleh, Iran, earthquake

Mapping the three-dimensional (3D) displacement fields associated with a variety of geological phenomena has been widely performed by exploiting synthetic aperture radar (SAR) imagery, as the result is important for providing insight into the formation mechanisms and potential risks of geological hazards. New-generation SAR sensors, namely ALOS-2 and Sentinel-1, can capture surface deformation with a high coherence in wide-swath mode, thereby providing outstanding across-track displacement accuracies; however, this improvement partially sacrifices the azimuth resolution, which affects the retrieval of 3D surface deformation fields. To explore the feasibility of generating 3D deformation maps with new SAR imagery, we collect two pairs of ALOS-2 ScanSAR and four pairs of Sentinel-1 Terrain Observation by Progressive Scans (TOPS) images for the 12 November 2017 Ezgeleh earthquake. Furthermore, the differential interferometric SAR (DInSAR), pixel offset tracking (POT), multiple-aperture InSAR (MAI), and burst-overlap interferometry (BOI) methods are used to measure the across- and along-track displacements. Compared with the POT and MAI methods, the integration of DInSAR and BOI measurements provides high-quality 3D deformation maps with an accuracy of 4 cm, which is four times and two times better than the accuracies of the POT and MAI methods integrated with DInSAR, respectively. In addition, a significant north–south displacement of 0.76 m is found in our 3D deformation results that was underestimated in the slip distribution model constrained with seismic waveforms or InSAR measurements. Our 3D deformation map of the 2017 Ezgeleh earthquake indicates a southwestward horizontal motion and an upward motion without any corresponding surface rupture that effectively match the behavior of a blind rupture along a northeast-dipping reverse fault. We conclude that combining BOI with DInSAR would provide a better 3D deformation field and should be applied to study future earthquakes.