多因子约束的自适应四叉树InSAR数据降采样方法

Multi-factor constrained downsampling method for adaptive quadtree InSAR data

摘要:

InSAR形变场的数据压缩有助于提高同震滑动分布模型和发震构造参数的反演效率。但如何在剔除噪声和冗余信息与保留更多全局性和局部性关键形变特征之间取得平衡对传统压缩方法是一个挑战。为此，本文提出一种基于方差、形变梯度和相干性多因子共同约束的InSAR同震形变数据降采样方法，选取三个不同复杂性的震例进行了验证。结果表明，该方法保持了较低的采样率和更好的鲁棒性，可有效地剔除低相干性点，对关键性形变细节保留度较高，且对形变的线性分布约束更强。

Abstract:

Down-sampling of InSAR displacement fields helps to remarkably improve inversion efficiency of inversion of coseismic slip distribution and fault geometry. However, how to keep balance between removing noise and redundant information and retaining more global and local key displacement information is a challenge to traditional down-sampling methods. For this reason, we here proposed a downsampling method constrained by multiple thresholds of variance, deformation gradient and coherence for coseismic deformation fields, which was verified with three strong events associated with complex fault systems. The result suggests that this method can effectively eliminate low coherence points, preserve critical deformation details, and has lower sampling rate, better robustness and stronger constraints on the linear distribution of deformation.