An Approach for the Semantic Enrichment of Sentinel-1 Imagery Suitable for Large-scale Analysis

Challenge:

Synthetic Aperture Radar (SAR) Earth observation (EO) satellites have several advantages over their optical counterparts, such as being able to observe the Earth's surface at night, and through a wide variety of weather conditions. However, due to the nature of their sensors and mechanisms of capture, the resultant imagery is often difficult to interpret and use in downstream analyses. Several approaches exist for the semantic enrichment of optical data, such as the Satellite Imager Automatic Mapper (SIAM*), which, coupled with their use in EO data cubes, can greatly improve accessibility and use of the original data. A system offering similar benefits for SAR EO data could be highly beneficial, especially considering the potential to complement optical data. Designing such a system to permit analyses across differing geographic areas globally presents an additional challenge which we have also attempted to address in this work.