<p>The PAZ SAR satellite, launched in 2018, routinely delivers X-band SAR (synthetic aperture radar) imagery in co-polarimetric HH and VV channels on a weekly basis. It has the potential to reveal surface elevation and deformations and to categorize scattering characteristics. Yet, few relevant experiments and studies have been carried out so far [1], probably due to the limited PAZ data availability to the public. Using a relatively small stack of 10 PAZ co-polarimetric SAR images, we investigate and demonstrate the applicability of PAZ co-polarimetric SAR imagery for monitoring surface deformation. Images were acquired between September 2019 and April 2020, covering the northern part of the Netherlands. This InSAR (interferometric SAR) time series of images allowed us to classify radar scatterers in terms of scattering mechanisms.</p><p>&#160;</p><p>A key linchpin in time series analysis for surface deformation monitoring is to identify reliable constantly coherent scatterers (CCS) and to maximize their number separately in the VV and HH channels. Sufficient and reliable CCS can facilitate spatio-temporal phase unwrapping, and map surface deformation evolution. A real-valued IRF (impulse response function) correlation method is suggested for CCS selection as it generates the CCS with exact radar location and maximum exclusion of incoherent scatterers and scatterers at the sidelobes. In this way it serves as an alternative to classical methods such as the normalized amplitude dispersion (NAD). The results of our study show that 34% CCS in the VV channel and 47% in the HH channel have an ensemble temporal coherence > 0.9 using the real-valued IRF correlation method, while 5% CCS in both the VV and the HH channel have an ensemble temporal coherence > 0.9 using the NAD method. Therefore, using the real-valued IRF correlation method we obtain better-quality results of the selected CCS. &#160;</p><p>&#160;</p><p>By using SAR images in both the VV and HH channels, co-polarimetric phase differencing (CPD) can be applied to classify the CCS into three classes: surface scattering, dihedral scattering and volume scattering. The results of our study show that by predefining an allowable noise range, in our study equal to 0.4, and using the temporal averaged CPD, we can achieve a reliable CPD-based classification. A higher percentage of CCS in the VV channel is classified as dihedral scatterers (24%), while a higher percentage of CCS in the HH channel is classified as surface scatterers (36%) and volume scatterers (47%).</p><p>&#160;</p><p>We conclude that PAZ co-polarimetric SAR imagery improves monitoring of surface deformation as compared to existing methods, and is suited to characterize radar scatterers.</p><p>&#160;</p><p>[1] Ling Chang and Alfred Stein (2020). Exploring PAZ co-polarimetric SAR data for surface movement mapping and scattering characterization. International Journal of Applied Earth Observation and Geoinformation. (https://doi.org/10.1016/j.jag.2020.102280)</p>
Prediction of categorized Sea Ice Concentration from Sentinel-1 SAR images based on a Fully Convolutional Network