GNSS-constrained InSAR correction for land subsidence mapping in Tianjin, China

View Online

Interferometric synthetic aperture radar (InSAR) is a primary geodetic technique for monitoring surface deformation, achieving high spatial resolution and extensive coverage. However, long-wavelength errors – such as orbital errors, atmospheric delays, and tidal loading – mask subtle deformation signals. Furthermore, reliance on a single reference station neglects plate motion gradients, causing time series displacement biases. These factors limit millimeter-level monitoring accuracy. Our study integrates InSAR and global navigation satellite system (GNSS) data, leveraging GNSS measurements to correct interferograms for long-wavelength errors and thereby enhance InSAR deformation accuracy. This integration also anchors InSAR-derived deformation to GNSS motions and helps establish a reference frame. Using Tianjin in northern China as our case study, we corrected long-wavelength errors in both ascending and descending Sentinel-1 interferograms with continuous GNSS observations. To evaluate the effectiveness of our long-wavelength error correction method, we compared the corrected interferograms against results derived from both the general atmospheric correction online service (GACOS) atmospheric correction and trend surface modeling. We conducted a systematic evaluation of correction accuracy and its impacts on four key results: phase changes, velocity estimates, cumulative displacement, and seasonal signals. Results show that the correction by using GNSS data improved accuracy in 84% of the 862 processed interferograms, while the performance of the correction depended on the baseline distance of the GNSS stations. The root mean square error (RMSE) of the time series displacement for the GNSS validation stations was maintained within the millimeter-level, improving the displacement accuracy by at least 80%. Regarding reference stability, the GNSS correction method proved to be more reliable than the single-reference approach. These results highlight the benefits of multiple GNSS stations in InSAR correction. Finally, our analysis of the fused vertical velocities, derived from both ascending and descending tracks, reveals a notable reduction in land subsidence across Tianjin since 2016, with subsidence rates and areas decreasing annually.

干涉合成孔径雷达（Interferometric Synthetic Aperture Radar, InSAR）是监测地表形变的重要大地测量技术，具有高空间分辨率和大范围覆盖等优势。然而，长波长误差——如轨道误差、大气延迟和潮汐负荷等——会掩盖微弱的形变信号。此外，依赖单一参考站会忽略板块运动梯度，从而导致时间序列位移偏差。这些因素限制了毫米级形变监测精度的实现。本研究融合 InSAR 与全球导航卫星系统（Global Navigation Satellite System, GNSS）数据，利用 GNSS 观测对干涉图中的长波长误差进行校正，从而提高 InSAR 形变监测精度。该融合方法还可将 InSAR 反演形变锚定至 GNSS 运动，并有助于建立统一参考框架。本研究以中国北方天津地区为例，利用连续 GNSS 观测对 Sentinel-1 升轨和降轨干涉图中的长波长误差进行了校正。为评估该长波长误差校正方法的有效性，本文将校正后的干涉图与通用大气校正在线服务（General Atmospheric Correction Online Service, GACOS）大气校正结果以及趋势面建模结果进行了对比。研究进一步系统评估了校正精度及其对四类关键结果的影响，包括相位变化、速度估计、累积位移和季节性信号。结果表明，在处理的 862 幅干涉图中，基于 GNSS 数据的校正在 84% 的干涉图中提高了精度，但其校正效果受 GNSS 站点基线距离的影响。GNSS 验证站的时间序列位移均方根误差（RMSE）保持在毫米级，位移精度至少提高了 80%。在参考稳定性方面，GNSS 校正方法相比单一参考点方法表现出更高的可靠性。上述结果表明，多 GNSS 站点在 InSAR 误差校正中具有明显优势。最后，基于升轨和降轨数据融合得到的垂向形变速率分析结果显示，2016 年以来天津地区地面沉降明显减缓，沉降速率和沉降面积均呈逐年下降趋势。