Research on GBSAR Deformation Extraction Method Based on Three-Threshold

Abstract

Ground-based synthetic aperture radar (GBSAR) is an effective means for regional, all-day, all-weather and real-time deformation monitoring and the detection of permanent scatterers (PS) is the key technology for GBSAR acquiring deformation. In order to effectively detect the PS point and analyze the disaster area data in real time, especially to solve the problem that the phase quality of the PS point changes dynamically with time when using the traditional PS point detection method, we present a deformation extraction method based on the three thresholds of coherence coefficient, amplitude dispersion and phase error. First, the method selects the PS candidate (PSC) points by using the average coherence coefficient and the amplitude dispersion threshold which considers the signal characterized by high signal-to-noise ratio (SNR) and stability. Then, we use the phase error threshold to perform secondary screening on the PSC points. In the deformation monitoring experiment of the landslide in Mao county, we compared and analyzed the actual effects of different PS point selection methods and illustrate the influence of different PS point selection effects on the deformation accuracy. The experimental results show that the three-threshold method for detecting PS point is more efficient than the single-threshold method and the double-threshold method, at the same time, it can reduce the deformation inversion error.