Abstract:

With the rapid development of interferometric synthetic aperture radar(InSAR)technique, massive and high-quality interferograms make the surface displacement monitoring over a larger area become available. And it has become a daily and professional work to acquire large-scale surface deformation in many countries or regions. However, the standard frame width of spaceborne SAR imagery is limited. Generally, the width of single SAR image is tens of kilometers or even hundreds of kilometers. As a result, large research area may not be covered by single track, such as interseismic displacements in Tibetan plateau and coseismic displacements in Wenchuan earthquake. It is necessary and crucial to integrate multi-track InSAR displacement on the purpose of acquiring large-scale surface deformation observations. This study focuses on the application requirements of large-scale interseismic deformation rate field reconstruction. We first generated a simulated dataset and quantitatively analyzed the error sources in integrating multi-track InSAR displacement rate products into a regional displacement map, including the inconsistency of reference point, side looking incident angle and other non-deformation signals(residual atmospheric errors, time difference of acquiring SAR images, etc.). Especially the influence of incident angle of different tracks was quantitatively explained and discussed. The deformation rate obtained by InSAR technology is relative to a specific reference point, which is a specific coherent target selected in the processing of InSAR data, and its active characteristics are often unknown. This makes the deformation rate field of different tracks have a systematic deviation, which needs to be compensated for integration. GPS displacement rate is usually applied to uniform reference datum. The imaging time of multi-track SAR images is different, and the length of time is also different, which makes the deformation rate in the overlapping area of adjacent tracks different. This is mainly caused by the nonlinear component of the deformation rate. When the time span and imaging time are not different, the images of observation time are generally not considered in interseismic deformation observation. Through the simulated experiment, it is found that the influence of incident angle must be considered to obtain high-precision and large-scale InSAR deformation rate field. And a second-order polynomial could be used to correct incident angle. Afterwards, we proposed a ratio method based on polynomial estimation to correct the incident angle. Additionally, the application scenario and effect of integration after transforming to ALD(Azimuth Look Direction)are discussed and analyzed for large-scale strike-slip faults. This study takes the southeastern Tibetan plateau as an experimental site. The integration of large-scale InSAR deformation rate field is conducted and analyzed. The GPS horizontal velocity field covered by InSAR data is used to calibrate the reference datum. In order to obtain the InSAR deformation corresponding to GPS station, we take the GPS point as the center with a radius of 5km, and use the inverse distance weighting method to obtain the InSAR deformation rate value. After the reference datum is corrected, the difference of InSAR deformation rate in overlapping areas is significantly reduced. The developed ratio method and conversion to ALD direction are used to integrate the InSAR displacement products reconstructed based on Sentinel-1 satellite data. The results show impacts of incident angle differences on the adjacent tracks and also prove the reliability of the strategies applied in this study. Under the condition of ensuring that the time coverage of multi-track SAR data is basically the same, the difference between the reference datum and the incident angle should be taken into consideration firstly to realize the fusion and integration of multi-track InSAR deformation rate field. And the developed ratio method and conversion to ALD direction are able to suppress or avoid the impact of incident angle for acquiring large-scale deformation field with high precision and high resolution. The ratio method based on polynomial estimation could correct the influence of incident angle. But the final result corrected by ratio method is located in the unilateral side view reference coordinate system, which is likely to cause far-field deformation and visual distortion. The strategy of conversion to ALD direction could also be applied for integration of large-scale InSAR deformation rate field, and is suitable for large-scale strike-slip fault areas dominated by horizontal displacement. The two strategies are of great significance to realize large-scale and high-precision InSAR deformation rate field reconstruction and they are successfully applied to integration of multi-track InSAR deformation products in the southeast of Tibetan plateau. And the strategy of conversion to ALD direction is slightly better than the ratio method, which shows that ALD transformation strategy is more suitable for southeastern Tibetan plateau. It shows that the actual displacements of the target area are mainly dominated by horizontal displacements, and the premise assumption of ALD conversion strategy is basically tenable. There is no unknown parameter estimation in the ALD conversion strategy, and it is simple and direct, while the ratio method requires polynomial fitting of the overlapping region, which is affected by the residual error in the deformation rate field.

Key words: temporal InSAR, GPS, multi-track integration, error analysis, incident angle

摘要：

随着合成孔径雷达干涉测量技术(Interferometric Synthetic Aperture Radar, InSAR)的飞速发展, 海量高质量的干涉图使得大面积地表形变监测成为可能。但星载SAR的标准景幅宽有限, 需要将多轨InSAR数据进行拼接来开展大范围地表形变的监测。聚焦于地震震间形变研究中广域形变场重建的应用需求, 文中首先基于模拟数据集定量分析了入射角对多轨道InSAR形变拼接的影响, 讨论了多轨道单一方向观测时InSAR形变场拼接中的主要误差来源; 提出了基于多项式估计的比值法修正入射角的方法; 面向大型走滑断裂, 讨论并分析了转换到地距向(Azimuth Look Direction, ALD)后进行拼接的应用场景和效果。最后, 以青藏高原东南部为实验区域, 利用GPS水平速度场进行参考基准校正, 分别使用比值法和转换到ALD方向对基于哨兵1号卫星重建的3个轨道InSAR数据进行拼接, 获得了大范围、 高精度的InSAR形变速率场。结果表明, 入射角的差异会造成相邻轨道同一区域InSAR震间形变速率场的差异, 当难以对InSAR形变场进行三维分解时, 文中提出的拼接策略可将多轨道的InSAR形变速率场的参考基准统一到同一空间参考框架下, 能够有效抑制入射角的影响, 实现广域InSAR震间形变速率场产品的拼接。

关键词: 时序InSAR, GPS, 多轨拼接, 误差分析, 入射角