Detection and correction of Azimuthal Stripes from L-band Synthetic Aperture Radar over the Low-latitude region using sliding-PCA technique

Kurzfassung

Synthetic Aperture Radar (SAR) signals operating at P-band and L-band frequencies over the low-latitude region are deeply affected with various impairments during the trans-ionospheric propagation [1]. Multiple undulations are visible on the SAR image with the pattern of stripes primarily caused due to the Ionospheric plasma turbulence aligned in the direction of the local geomagnetic field. Accordingly, the SAR signal acquisition will be undergoing through the deterministic, temporal and random ionospheric structures. In addition, with the two-way trans-ionospheric signal propagation the effect should be calibrated in the SAR system. Especially, the geometric calibration could be severely affected by the Ionospheric impact on SAR image. One of the predominant artifacts on SAR image could be azimuthal stripes in SAR amplitude occurs due to constructive and destructive phase interference due to small-scale ionospheric irregularities. The scale-size and elongation of such equatorial plasma bubbles (EPBs) or the plasma density irregularities provides denser to coarser wave patterns of azimuthal stripes on the SAR image. Several researchers proposed the correction techniques to mitigate the azimuthal stripes on the received SAR signal amplitude. Frequency domain filtering approaches using Fourier analysis and Gaussian Notch Filtering (GNF) are proven to be efficient by applying the notch filters for the corresponding ionospheric scintillation wave components in the spatial wave-vector space [2]. The efficiency of this method depends on the proper selection of mean and standard deviation of the Gaussian notch filter wave components. We present a new approach using the sliding principal component analysis (sliding-PCA) to detect and correct the stripes on the SAR image. Proposed methodology is tested on the ALOS-2/PALSAR L-band SAR data collected on 15 and 29 April 2023 centered at geographic longitude: 99.899° E and geographic latitude: 14.621° N with severe ionospheric scintillation impacts acquired over the Thailand region. An attempt is made to estimate the amplitude scintillation index (S4) from the SAR image and validated with the co-located GNSS measurements. Effectiveness of the proposed sliding-PCA filtering technique is validated using the multi-structural similarity index (MSSI) determined the usefulness in the calibration for the L-band SAR images.