The atmospheric phase, which is the sum of vertical stratification and turbulent atmospheric phase, is a major challenge currently faced by small baseline subset interferometric synthetic aperture radar (SBAS-InSAR) measurements. Many previous studies have demonstrated that the former can be separated from the interferogram by establishing a functional model between it and the topography. Due to the high variability of the turbulent atmospheric phase (TAP) in the space and time domains, however, the TAP is difficult to model and remove. Recently, many stochastic models have been developed to reduce the influence of the TAP in SBAS-InSAR. To avoid the rank deficient in stochastic model method, we present a correction method using network-based variance estimation, interferogram stacking and ordinary kriging interpolation (NIO). There are three main steps in the proposed algorithm to ensure the accuracy of the correction result: (1) adaptively identify and select sufficient good-quality interferograms that contain less turbulent atmospheric noise to participate in deformation calculation; (2) further select the short temporal baseline interferogram and mask the corresponding deformation location to avoid the effect of deformation; and 3) take advantage of ordinary kriging interpolation to reduce the effects of TAP from the selected good-quality interferograms. The performance of the proposed method has been validated with a set of simulations and real Sentinel-1A SAR data in Southern California, USA.

大气相位是垂直分层和湍流大气相位的总和，是小基线子集干涉合成孔径雷达（SBAS-InSAR）测量目前面临的主要挑战。先前的许多研究表明，通过在干涉图和地形之间建立功能模型，可以将前者与干涉图分开。然而，由于湍流大气阶段（TAP）在空间和时间域中的高度可变性，TAP 难以建模和去除。最近，已经开发了许多随机模型来减少 TAP 对 SBAS-InSAR 的影响。为了避免随机模型方法中的秩缺陷，我们提出了一种使用基于网络的方差估计、干涉图叠加和普通克里金插值（NIO）的校正方法。为了保证校正结果的准确性，该算法主要分为三个步骤：（1）自适应识别并选择足够多的、包含较少湍流大气噪声的优质干涉图参与变形计算； (2)进一步选择短时间基线干涉图并屏蔽相应的变形位置，以避免变形的影响； 3) 利用普通克里金插值法来减少所选高质量干涉图中 TAP 的影响。该方法的性能已通过美国南加州的一组模拟和真实的 Sentinel-1A SAR 数据得到验证。