Potential of Using Sentinel-1 Data to Distinguish Targets in Remote Sensing Images

Abstract

Copernicus is the World’s largest single Earth Observation (EO) programme, whose satellite constellations are planned to be launched between 2014 and 2025. Among the constellations, Sentinel-1 (S-1) is a C-band SAR able to support land cover mapping. Although optical data are commonly used for land cover monitoring, the low availability of cloud-free scenes along the year hinders the mapping process. In such a way, S-1 presents an important source of data, able of providing all-weather and day-and-night imagery of EO. In this study, we investigate the potential of using S-1 data to distinguish targets in Remote Sensing images in three different Brazilian biomes, Amazon, Cerrado, and Atlantic Forest. Based on that, we proposed a methodology to classify SAR images, which was validated considering a different area from the ones used for sampling purposes. The results showed that through S-1 data, it is possible to detect mainly water and urban area targets, with overall accuracy of 0.90, evidencing that our approach is reproducible in other regions.