Patch-Like Reduction (PLR): A SAR Offset Tracking amplitude filter for deformation monitoring

https://doi.org/10.1016/j.jag.2022.102976Get rights and content
Under a Creative Commons license
open access

Highlights

  • The causes of Patch Like errors in Offset Tracking results have been investigated.

  • An amplitude filter (PLR) has been proposed to reduce PLs.

  • The filter threshold is automatically defined from images amplitude distributions.

  • The filter improves OT results by reducing the changes of PL appearance.

  • Template sizes and tapering windows have been optimized in terms of resolution.

Abstract

As complementary to Synthetic Aperture Radar (SAR) Differential Interferometry (DInSAR), SAR Offset Tracking (OT) is an efficient tool for large ground deformation monitoring in situations when DInSAR cannot work. However, SAR images are affected by speckle noise and some strong point-like scatters which can cause what is known as Patch Like (PL), a kind of errors that can be seen as homogeneous patches of almost constant deformation in the results. These errors are clearly visible in the results as non-consistent deformations along time, but they are difficult to detect with the traditional metrics that evaluate the cross-correlation results, like the Signal to Noise Ratio (SNR). This paper addresses this problem and proposes a simple amplitude filter to reduce PL named as Patch Like Reduction (PLR). The main idea is to find a sensor and scene independent threshold to remove the high amplitude pixels prone to cause PL. Five different SAR data sets and in-field GPS measurements are used to determine the optimal threshold and evaluate the performance of the proposed method. The results show that PL effects can be reduced with the proposed amplitude filter. The processing parameters of the improved OT processing chain are optimized as well to preserve the results resolution as much as possible.

Keywords

Offset tracking
Ground deformation
Synthetic aperture radar (SAR)
Patch like
Amplitude filter
Window function
Template size

Data availability

The authors do not have permission to share data.

Cited by (0)