Key technical issues in seismic crustal deformation measurement by InSAR

Abstract

Interferometric Synthetic Aperture Radar (InSAR) has been proved to be a high-precision geodetic approach for monitoring the crustal deformation of the earth. Comparison with other methods, such as leveling, GPS, Very-Long-Baseline Interferometry (VLBI), Satellite Laser Ranging (SLR) and so on, InSAR has the remarkable advantages of continuous large-area coverage and centimeter to millimeter-level measurement accuracy, its measuring results are the significant geophysical parameters needed by seismic research, thereby becoming an important means of seismic crustal deformation monitoring. The occurrence and development of earthquakes have periodic characteristics. A complete seismic cycle can be divided into four phases: pre-seismic, co-seismic, post-seismic and inter-seismic. Different seismic processes have their own crustal deformation characteristics, so we should to adopt different methods for the deformation monitoring at different stages. At present, D-InSAR technology is mainly used to monitor the co-seismic deformation, while time-series analysis of InSAR is in the inter-seismic deformation monitoring. In recent years, some new satellite platforms, namely, Sentinel-1A/1B RADARSAT-2, ALOS-2, TerraSAR/TanDEM-X and COSMO-SkyMed, have enhanced the ability to obtain SAR data globally with short revisit cycles and possibility for monitoring crustal deformations worldwide and routinely. In the meantime, the number of InSAR users, including governments, research institutes, and commercial companies, is expanding year by year, then the demands to improve the accuracy and reliability of InSAR results are also increasing. This paper first summarizes the landmark works and up-to-date research status of InSAR technology in seismic crustal deformation monitoring, then focuses on major limitations hindering the technique deeper applications, such as Atmospheric Phase Screen (APS), Line-Of-Sight (LOS) ambiguity, and imaged swath width. The existing methods to remove APS have been summarized, and an emerging method based on Numerical Weather Prediction (NWP) model has been discussed especially. The problems on swath width limitation and LOS ambiguity have been analyzed and the possible solutions are introduced. InSAR time-series analysis is an advanced method for monitoring inter-seismic and post-seismic displacement. In the paper, the most important four methods of InSAR time-series, which are PS-InSAR, SBAS, StaMPS, and SqueeSAR respectively, are described in detail. The paper concludes with a discussion of the key technical issues on InSAR applications and the associated ways toward to the solutions. The conclusions are given as follows: (1) APS is a main source of error in InSAR processing, which could be eliminated or mitigated by APS correction with the output from NWP models. (2) One-dimensional measurement along the LOS direction has greatly limited the capability of InSAR technique in the investigation of crustal deformations which demands three dimensional deformation components. To obtain accurate 3-D surface deformation not only need to incorporate descending InSAR result together with ascending one, or with Multi-Aperture Interferometry (MAI), but also SAR satellite in large oblique orbit should be considered. (3) High-coherence target selection is an important prerequisite for InSAR time series analysis. The problems, such as unevenly distribution of PS and low coherence of the natural surface, are main reasons for poor performance of InSAR time-series analysis application in Earthquake research. Reasonable choice of the thresholds and optimal high-coherence target selection strategies can improve the accuracy of InSAR results. (4) Wide area mapping is urgently needed in the crustal deformation monitoring, since the coverage of the great earthquakes faults usually extended over several hundreds or even up to thousands kilometers. Strategies to merge multi-track InSAR results together with ScanSAR interferometry and its time series analysis should be took into consideration in future. (5) Time-series analysis algorithms must be adapted to incorporate each new image in an efficient and optimal manner without starting the processing from scratch. New approaches should be proposed as efficient processing schemes to exploit the unprecedented Big Data for high-precision near-real-time processing. (6) Now the validation of InSAR outputs or its accuracy estimation relies heavily on GPS, leveling, and other external data. Here, we propose an idea to introduce the " Totally Quality Control" into InSAR processing chain that goes through every step of InSAR time series processing to indentify the possible artifacts in the processing and correcting them to ensure the quality of the outputs.