Statistical Theory for the Detection of Persistent Scatterers in Insar Imagery
Author | : Stacey Amy Huang |
Publisher | : |
Total Pages | : |
Release | : 2021 |
ISBN-10 | : OCLC:1266221917 |
ISBN-13 | : |
Rating | : 4/5 ( Downloads) |
Download or read book Statistical Theory for the Detection of Persistent Scatterers in Insar Imagery written by Stacey Amy Huang and published by . This book was released on 2021 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: Interferometric Synthetic Aperture Radar (InSAR) is a powerful remote sensing technique for observing subtle deformation of the Earth's surface over time through multiple observations of the same ground area. Because radar backscatter depends on wavelength-scale properties of surfaces, traditional InSAR methods can fail over naturally changing terrain. The persistent scatterer InSAR (PS-InSAR) technique is one important extension for time-series analysis which identifies and utilizes only the most reliable points in InSAR images for analysis. PS-InSAR has been successfully applied to detect mm-level deformation associated with natural hazards such as earthquakes, volcanoes, and landslides. To date, however, the implementation of PS-InSAR has not been fully optimized, which can limit its utility in challenging mixed-terrain regions. In this thesis, we show that these techniques can be further optimized by characterizing the statistics of PS and developing a statistical framework for applying PS-InSAR techniques. There are three major parts to this work. First, we analyze PS density for different terrain types and image resolution and present a model for predicting the change in PS density, which adheres to empirical results within 50% error and closer for points that form the desired network for PS. Second, we characterize the probability distribution functions (PDFs) of the backscatter from PS and non-PS (clutter) and find that both are highly non-Gaussian over a variety of bandwidths and wavelengths. Finally, we demonstrate a novel maximum likelihood PS detector based on these non-Gaussian models. We show results from the improved detector over Hawaii's Kilauea Volcano and California's Central Valley. In both areas, the non-Gaussian detector finds many more PS than in the existing detector, which leads to a more complete map of deformation. Further, we find that the retrieved deformation time-series is consistent with that measured with three other methods: the existing maximum likelihood Gaussian detector, the small baseline subset (SBAS) InSAR method, and GPS.