Abstract While the quality and availability of Interferometric Synthetic Aperture Radar (InSAR) data has dramatically improved in recent years, InSAR analysis and interpretation remain challenging in actively deforming regions with extensive agricultural activities, where vegetation changes and soil moisture variability can degrade data quality or introduce confounding signals. Here we use Sentinel-1 satellite imagery for the Imperial Valley, California, over the period of 2015–2019 to explore how factors specific to land surface changes may impact InSAR time series and to resolve time-varying deformation due to tectonic processes and geothermal energy production. We examine the temporal variability of data quality, via interferometric phase coherence, at high spatial resolution, taking into account the observation that some agricultural fields lie fallow for long time intervals punctuated by periods of cultivation. This strategy allows us to better distinguish signals and noise associated with agricultural activities, shoreline changes, or surface soil conditions. A series of masking, interpolation, and filtering steps facilitate phase unwrapping, and the unwrapped, unfiltered product is then recovered, reducing artifacts from spatial filtering. We adopt model-based tropospheric corrections to improve time series results, particularly in regions with high topographic relief, along with the use of distributed reference points to render a more uniform error structure. We validate InSAR observations with continuous GPS where available and find that the estimates of average line-of-sight (LOS) velocity over the valley from InSAR and GPS agree to ~3 mm/yr in areas with good data coverage. Discrepancies between the two estimates often exist in areas with lower InSAR data quality; in better-constrained areas, they appear to reveal signals attributable to surficial processes occurring in the uppermost soil layers. We observe a diverse suite of natural signals over multiple spatial scales, including steady interseismic deformation, seasonal lake-level-modulated signals at the southeastern Salton Sea shore, and transient slow slip on the Superstition Hills fault. In addition, we observe complex deformation at four geothermal fields within the valley. Extensive subsidence at the Salton Sea geothermal field suggests spatial overlap of anthropogenic and tectonic deformation, interspersed with potential surficial signals. Geothermal sites at East Mesa, North Brawley, and Heber exhibit smaller-amplitude, more localized deformation, often with nonlinear temporal trends. Our analysis demonstrates the need to assess whether InSAR signals result from surficial changes or deeper sources, and produces robust ground deformation data in support of efforts to study subsurface processes, manage geothermal operations, and improve hazard assessments.

中文翻译：

---

加利福尼亚州帝王谷存在持续地表变化的相干引导 InSAR 变形分析

摘要 虽然近年来干涉合成孔径雷达 (InSAR) 数据的质量和可用性有了显着提高，但 InSAR 分析和解释在农业活动广泛的活跃变形地区仍然具有挑战性，在这些地区，植被变化和土壤水分变异会降低数据质量或引入混淆信号。在这里，我们使用 2015 年至 2019 年期间加利福尼亚州帝王谷的 Sentinel-1 卫星图像来探索特定于地表变化的因素如何影响 InSAR 时间序列，并解决由于构造过程和地热能引起的时变变形生产。我们通过干涉相位相干，在高空间分辨率下检查数据质量的时间变化，考虑到观察到一些农田长时间休耕，中间有耕作期。这种策略使我们能够更好地区分与农业活动、海岸线变化或表层土壤条件相关的信号和噪声。一系列掩蔽、插值和滤波步骤促进了相位解缠，然后恢复解缠、未滤波的产物，减少空间滤波的伪影。我们采用基于模型的对流层校正来改善时间序列结果，特别是在地形起伏较大的地区，同时使用分布式参考点来呈现更均匀的误差结构。我们在可用的情况下使用连续 GPS 验证 InSAR 观测，并发现 InSAR 和 GPS 对山谷上的平均视线 (LOS) 速度的估计在数据覆盖良好的区域中与大约 3 毫米/年一致。在 InSAR 数据质量较低的地区，这两种估计值之间往往存在差异；在约束较好的地区，它们似乎揭示了可归因于发生在最上层土壤中的表层过程的信号。我们在多个空间尺度上观察到一系列不同的自然信号，包括稳定的地震间变形、索尔顿海岸东南部的季节性湖泊水位调制信号，以及迷信山断层上的瞬时缓慢滑动。此外，我们还观察到山谷内四个地热场的复杂变形。索尔顿海地热田的广泛沉降表明人为和构造变形的空间重叠，并穿插着潜在的地表信号。East Mesa、North Brawley 和 Heber 的地热站点表现出较小的幅度、更多的局部变形，通常具有非线性时间趋势。我们的分析表明，需要评估 InSAR 信号是来自地表变化还是更深的源，并生成可靠的地面变形数据，以支持研究地下过程、管理地热作业和改进灾害评估的努力。